1. Introduce yourself/your company and describe your role. What problem were you compelled to solve?
I’m Pek Pongpaet, CEO of Impekable. We’re a Silicon Valley-based AI implementation consultancy that’s been around for over a decade — we’ve delivered for Nike, Google, Adobe, and Netgear.
But the work I’m most focused on right now is AI voice and contact center automation. Here’s why.
Every week I talk to executives running businesses with 50, 100, 200-person call centers. They’ve all seen the demos. They’ve all sat through the pitch from some AI vendor showing a slick prototype that sounds amazing in a conference room. And then they ask the question nobody at those demos wants to answer: “Okay, but how does this actually work with our Salesforce? Our Twilio phone system? Our compliance requirements? Our agents who’ve been doing this for 15 years?”
That’s the gap we fill. We don’t build AI models. We take the best AI voice technology on the market — ElevenLabs, Google Contact Center AI, VAPI, Twilio Conversational AI, Decagon — and we wire it into the CRMs, telephony stacks, and workflows our clients already run. We turn promising pilots into production systems that handle real call volume, real edge cases, and real customers.
The problem I was compelled to solve is simple: there are too many impressive AI demos and not enough working systems. Executives are stuck in a loop — they see the potential, they run a pilot, it works in a controlled environment, and then it dies the moment it hits their actual infrastructure. We break that loop.\
2. What is your core operating/business model — in-house team, external partners, or hybrid?
Hybrid, and deliberately so.
We have a core team that handles architecture, product management, and client strategy — that’s the brain trust, and it doesn’t rotate. When you engage Impekable, either I or one of my GMs is your strategic point of contact. You’re not getting handed off to a junior PM after the sales call.
For engineering, we scale up or down based on the engagement. Some projects need a focused two-person team for five weeks. Others need a larger squad for a multi-phase rollout. We staff based on what the project actually requires, not based on who’s sitting on the bench.
We’re not just a lean operating team — we’re a lean company. There’s no bloated middle management layer, no sprawling departments burning overhead. That matters to our clients because it means you’re not paying for organizational weight you don’t need. Every dollar of your engagement goes toward people doing actual work
on your project, and you’re not getting a B-team because the A-team is busy elsewhere. Every engagement gets senior attention from start to finish.
3. How do you differentiate yourself/your company in a crowded market?
Three things, and they’re all hard to replicate.
First, our partnerships aren’t logos on a slide — they’re working relationships. We’re an official U.S. Partner for ElevenLabs, and we work closely with their sales team and co-implement alongside their customer success managers. We have a 10+ year partnership with Twilio covering voice, SIP trunking, SMS, omnichannel messaging, and Flex contact centers. We’re Google Cloud Partners focused on AI — including Vertex AI and Contact Center AI (CCAI). We also partner with VAPI, AWS, and GCP. These aren’t badges we collected — they’re ecosystems we operate in daily.
Second, we’re not generalists pretending to know AI, and we’re not AI researchers pretending to know enterprise systems. We sit at the intersection. We understand how a COO thinks about call center economics, how a CTO thinks about system architecture, and how an agent on the floor actually handles a call. That translation layer — between business reality and technical possibility — is what most consultancies can’t do.
Third, speed. We’ve shipped enough Twilio integrations, Stripe payment flows, and AI voice agents that we’re not figuring things out on your dime. A typical MVP build runs 3-5 weeks. We can give a fixed price and mean it — no padding, no surprise change orders — because we’ve built the core components before and we know what it actually costs.
4. Which industries or sectors do you primarily serve, and how has that focus evolved over time?
Today, our deepest expertise is in five verticals: call center automation, mortgage and financial services, healthcare, debt collection, and retail outbound sales.
The thread connecting all of them is the same: these are industries where the phone call is still the highest-value moment in the customer journey, and where the cost of handling those calls is eating into margins.
We started as a broader digital product studio — apps, platforms, web experiences for enterprise clients. Over the last several years, we’ve gone deep on voice AI and contact center technology because that’s where the biggest transformation is happening. When you see a company spending $15-20 per qualified transfer and you can cut that by 60-70% with an AI voice agent that handles intake, qualification, and routing — the ROI conversation is straightforward.
The evolution was organic. We built the entire call center application for a large loan consolidation company on Twilio — that led to more financial services work. Our ElevenLabs partnership brought us healthcare clients running 10,000+ outbound calls daily. Each vertical taught us the compliance nuances, the workflow patterns, and the edge cases specific to that industry. Now we don’t have to learn those lessons on a new client’s budget.
5. What services or solutions are most in demand from clients?
Three things dominate right now.
The first is AI voice agent implementation. A client comes to us — usually after they’ve seen an ElevenLabs or VAPI demo — and says, “We want this, but connected to our CRM, our phone system, and our compliance workflow.” We handle the full build: conversational flow design, telephony integration via Twilio or SIP trunking, real-time API tool calls for data lookups and eligibility checks, agent handoff logic, and ongoing optimization through transcript reviews and prompt tuning.
The second is contact center modernization. This is the executive who’s staring at a legacy call center — maybe running on an outdated platform, maybe cobbled together from three different tools — and knows they need to rebuild before they can even think about AI. We’ve done this end-to-end, from Twilio Flex omnichannel implementations to custom agent dashboards to queue management and real-time call routing.
The third — and this one’s growing fast — is fractional CTO and technology advisory work. An executive has budget, has buy-in from the board, maybe even has a vendor selected. But they don’t have anyone internally who can tell them whether the architecture makes sense, whether the vendor’s promises are realistic, or whether the integration plan will actually work. We come in as the trusted technology advisor who speaks both languages — business outcomes and system architecture. Sometimes that turns into a full build. Sometimes it’s a two-week assessment that saves them from a six-figure mistake.
6. How do you personally stay ahead of industry shifts?
I’m going to give you the honest answer, not the polished one.
The biggest advantage I have is that I’m in the room when these technologies ship. As an ElevenLabs U.S. Partner, I’m not reading about their new features in a blog post — I’m seeing what’s coming, testing pre-release capabilities, and understanding how new models will affect existing deployments. When you’re actively implementing this technology for clients every week, you’re not theorizing about what’s possible — you’re learning what actually works in production.
Beyond that, I use AI tools extensively in my own work. I use Claude daily for strategic thinking, content development, and synthesizing complex project requirements. It’s not just a research tool — it’s how I stress-test ideas, draft proposals, and think through architecture decisions before they hit a whiteboard. When I tell a client that AI can transform their workflow, it’s because I’ve already transformed my own.
The last piece is simply talking to operators. Not vendors, not analysts — the people running call centers, managing agent teams, watching their cost-per-call reports. Their problems don’t change as fast as the technology does, and understanding the problem deeply is more valuable than chasing every new tool.
7. Do you have repeat clients? What strategies contribute to that loyalty?
Yes, and the strategy is simple: we don’t treat engagements as projects. We treat them as relationships.
Our approach is to be the trusted advisor. We’re constantly looking for ways to help clients run more efficiently, save money, and reclaim time. Every recommendation we make is tied to a business outcome — not “here’s what we can build” but “here’s what this will improve, here’s how this saves money, and here’s how this makes you more money.” When you show up with that mindset, clients stop thinking of you as a vendor and start thinking of you as part of their team.
And we don’t disappear after launch. The work is never done — there’s always more to improve. For our AI voice clients, we do weekly transcript reviews, prompt tuning, and workflow adjustments. That ongoing involvement means we catch issues early, optimize continuously, and build institutional knowledge about the client’s business that no one else has.
If I’m being candid, the real loyalty driver is this: when something goes wrong — and in production systems, something always goes wrong — we’re the team that picks up the phone. Clients remember who was there when the integration broke, not who had the best slide deck.
8. How do you measure client satisfaction and success beyond the deliverables?
We measure it the way a business operator would, not the way a project manager would.
The first thing I look at is whether the system is actually running in production. Not whether we delivered it on time — whether it’s actually handling live calls, processing real transactions, routing real customers. I’ve seen plenty of agencies deliver a project “on spec” that never makes it to production because nobody thought about the last mile. We measure success by whether the thing works in the real world, not whether the Jira board is green.
The second is business metrics. For our AI voice implementations, we track cost-per-transfer, automation rates, call completion rates, and agent handoff quality. For a recent healthcare client, we projected 450% ROI with $675K net savings in Year 1 — and we structured the engagement so we’d be accountable to those numbers, not just the code delivery.
The third — and this one matters more than people think — is whether the client’s team can actually operate the system after we hand it off. We build with documentation, we give teams access from day one, and we design for maintainability. A recent client’s technical co-founder got GitHub access from the start so he could learn the codebase as we built it. That’s not a nice-to-have — it’s how you build a product someone actually owns.
9. What kind of post-project support do you provide?
We offer ongoing support retainers, and for AI voice clients, this is where a lot of the value lives.
An AI voice agent isn’t a website — you don’t launch it and walk away. The real optimization happens in the weeks and months after deployment. We do weekly transcript reviews to identify where calls are failing, where prompts need tuning, and where new edge cases are appearing. We adjust conversational flows, update tool call logic, and refine escalation paths based on real data.
For non-AI projects, we provide standard maintenance and support — bug fixes, infrastructure monitoring, and iterative improvements as the client’s business evolves.
The key point: post-project support isn’t an afterthought or an upsell. We scope it into the engagement from the start because we know from experience that the gap between “launched” and “working well” is where most projects die.
10. How do you structure pricing and billing?
We offer two models, and the choice depends on how much certainty the client needs.
The first is fixed price, fixed scope. You tell us what you need, we scope it, and we give you a number. No surprises, no change orders. We can quote a fixed price confidently because it’s based on comparable projects we’ve already delivered — not an estimate built on hope.
The second is flexible scope, where we work within a budget range. This works well for clients who know they need to move forward but don’t have a fully defined picture yet — they need to start somewhere and refine as they go. We set a range, prioritize together, and adapt as the engagement reveals what matters most.
For ongoing AI voice optimization, we offer monthly retainers that cover transcript reviews, prompt tuning, and workflow adjustments. The economics are straightforward: you’re paying for continuous improvement of a system that’s saving you multiples of our fee in reduced call center costs.
We don’t do hourly billing. It creates the wrong incentives for everyone.
11. What is the typical price range for projects, and how do you balance affordability with value?
Every engagement is different, and we scope based on the specific business problem — not a rate card.
What I can tell you is how we think about value, because it’s the same way our clients think about it: what does it cost to not do this?
If your call center is spending $15-20 per qualified transfer and an AI voice implementation cuts that by 40- 60%, the system pays for itself quickly — often within the first few months. We’ve seen clients project six figure annual savings from a single implementation. At that point, the question isn’t whether you can afford to build it — it’s whether you can afford to wait.
We also move faster than traditional agencies because we’ve built the core components before. The same platform another shop might quote six months and a much larger budget for, we can deliver in weeks — not because we cut corners, but because we’re not learning on your dime. Efficiency is our pricing advantage.
The best way to get a real number is to have a conversation about your specific situation. We’ll be direct about what it costs and what the expected return looks like. No ambiguity, no bait-and-switch.
12. Have you turned down projects? What are your minimum requirements for the right fit?
Yes, and it’s one of the best decisions we make regularly.
We turn down projects for three reasons. First, if the client doesn’t have a clear business outcome in mind. “We want AI” is not a project — it’s a wish. If we can’t tie the engagement to a specific metric — calls automated, cost reduced, revenue generated, time saved — we’re probably not the right fit.
Second, if the technical environment is so fragmented or undocumented that we’d spend more time untangling legacy decisions than building forward. We’ll sometimes recommend a smaller assessment engagement first to map the landscape before committing to a full build.
Third, if the budget doesn’t match the ambition. We’d rather have an honest conversation about what’s achievable at a given price point than over-promise and under-deliver. Sometimes that means recommending a smaller MVP scope. Sometimes it means telling a prospect they need a different type of partner.
Our minimum requirement is straightforward: the client needs a real business problem, a decision-maker at the table, and a willingness to move at the speed the market demands. If all three are present, we can do exceptional work together.
13. What key challenges have you faced, and how did you overcome them?
The biggest challenge we’ve faced is one every implementation consultancy hits eventually: the gap between what AI demos promise and what enterprise environments can actually support.
Early in our AI voice work, we’d see clients get excited by a demo, commit to a deployment, and then stall when they realized their CRM couldn’t handle real-time API calls, their telephony stack needed upgrading, or their compliance team hadn’t been consulted. The project would drag, the budget would swell, and everyone would blame the technology when the real issue was infrastructure readiness.
We solved this by building assessment into the front end of every engagement. Before we write a line of code, we map the client’s existing systems, identify integration gaps, and surface the non-obvious blockers — compliance requirements, data quality issues, agent training needs. It adds a week to the timeline and saves months of rework.
The second challenge has been hiring. Finding engineers who understand both modern AI tooling and legacy enterprise systems is genuinely difficult. Most developers want to work on greenfield projects, not figure out how to wire an ElevenLabs voice agent into a 10-year-old Zoho instance. We’ve built a team that actually enjoys that translation work, and it’s our biggest competitive advantage.
14. How do you foster innovation and adapt to emerging trends?
Innovation in our world doesn’t look like a hackathon or an R&D lab. It looks like deploying a new capability in a live client environment and seeing whether it holds up.
Our ElevenLabs partnership is the best example. As their U.S. Partner, we’re often among the first to implement new voice AI capabilities in production. When they release a new model or feature, we’re not evaluating it in a sandbox — we’re testing it against real call transcripts, real customer interactions, and real business metrics. That feedback loop — deploy, measure, iterate — is faster and more honest than any innovation program.
We also invest time in what I’d call “cross-pollination.” What we learn optimizing AI voice agents for healthcare lead qualification directly informs how we approach debt collection outreach or retail outbound sales. The industries are different, but the patterns — intake, qualification, routing, escalation — are remarkably similar. Every client engagement makes us better at the next one.
Within my own workflow, I use Claude as a daily thinking tool — drafting proposals, stress-testing positioning, synthesizing complex requirements. It’s changed how I work, and it gives me firsthand experience with the AI assisted workflows we’re building for clients.
15. What role does culture play in your success, and how do you build it?
Culture in a consultancy comes down to one thing: do you care more about shipping impressive work or about solving the client’s actual problem?
Those sound like the same thing, but they’re not. I’ve worked with plenty of talented engineers who wanted to build the most elegant system and plenty of project managers who wanted to run the most organized process. Neither of those matter if the client’s call center still can’t handle Monday morning volume.
We hire for people who get satisfaction from making something work in the real world — the messy, complicated, never-quite-as-clean-as-the-diagram real world. The engineer who’s excited about wiring an AI voice agent into a 10-year-old CRM. The designer who obsesses over whether a 65-year-old can navigate a mobile interface without calling their kid for help. The project lead who treats a client’s deadline like their own mortgage payment.
The other cultural element is transparency. We don’t hide problems. If an integration is more complex than we estimated, the client hears about it the same day — along with options and trade-offs. That honesty has cost us in the short term occasionally, but it’s the foundation of every long-term client relationship we have.
16. Where do you see your company/field in the next 5–10 years? What are your boldest long-term goals?
In 5 years, every company with a call center will have AI handling the majority of first-contact interactions. That’s not a prediction — it’s already happening. The question is who helps them get there without wrecking their operations in the process.
The companies that win won’t be the ones with the best AI model — that’s commoditizing fast. They’ll be the ones who can integrate that model into the hundred other systems a business runs on. CRM, telephony, billing, compliance, scheduling, escalation workflows — the plumbing is where value lives, and it’s where most deployments fail.
Impekable’s goal is to be the implementation partner that enterprises default to when they’re ready to move from AI pilot to AI production. Not the vendor who sold them the model. Not the consulting firm who wrote the strategy deck. The team that actually made it work.
Boldest long-term goal? I want Impekable to be to AI voice implementation what Accenture is to ERP — the name that comes up automatically when the project is too important to get wrong. But at our scale, with our speed, and without the bureaucracy.
17. How has your leadership style evolved throughout your career?
Early on, I was the person who wanted to be in every decision. When you’re building a company from scratch and your name is on the work, it’s hard to let go. Every design review, every architecture call, every client conversation — I was there.
That doesn’t scale, and more importantly, it doesn’t produce the best outcomes. The shift happened when I realized that my job wasn’t to have the best answer in the room — it was to make sure the right people were in the room and that they had the context to make good decisions.
Now my role is primarily strategic: scoping engagements, managing client relationships at the executive level, and making sure our technical approach aligns with the client’s business reality. I’m the person who sits across from a COO and translates “we need to reduce cost-per-call by 40%” into a specific technical architecture — and then trusts my team to build it.
The biggest evolution has been learning to be the “translator.” Most AI conversations fail because the technical people can’t explain business impact and the business people can’t evaluate technical claims. I’ve spent a decade learning both languages, and that’s where I add the most value now.
18. What emerging technologies or market shifts are you most excited about?
AI voice agents are the obvious one, but I want to be specific about why.
The shift that excites me isn’t that AI can talk — it’s that AI can now handle the boring, repetitive, high-volume interactions that burn out human agents and drain call center budgets. Intake calls. Qualification calls. Appointment scheduling. Payment reminders. These aren’t complex conversations, but they represent 60-70% of call volume for most businesses. When you automate those at $0.30 per call instead of $8-15 per call, you’re not just saving money — you’re freeing your best agents to handle the calls that actually require human judgment.
The second shift is the convergence of voice AI with real-time data. Our implementations aren’t just running scripts — they’re making API calls during the conversation. Checking eligibility in a database. Looking up account status in a CRM. Pulling lead scores from a marketing platform. The voice agent becomes a front-end for your entire business system, not just a phone tree replacement.
The third is the democratization of enterprise AI. Two years ago, building a production AI voice system required a team of ML engineers and six months of development. Now, with platforms like ElevenLabs and VAPI, we can deploy production-ready voice agents in 3-6 weeks. That opens the door for mid-market companies — $50M to $500M revenue — that couldn’t have afforded this 24 months ago.
19. What advice would you give to aspiring leaders or founders?
Build something that works before you build something that scales.
I see too many founders — especially in AI right now — chasing scale before they’ve proven the unit economics work for a single customer. Your first deployment, your first client, your first 1,000 calls matter more than your pitch deck’s TAM slide.
Second, learn both languages. If you’re technical, learn to read a P&L and talk to a CFO. If you’re a business operator, learn enough about the technology to know when a vendor is overselling and when an engineer is over-building. The leaders who create the most value are the ones who can sit in both rooms and be credible.
Third, pick partners based on what they’ve built, not what they say they can build. When I talk to executives evaluating AI vendors and implementation partners, I always ask: can they show you a working system in a comparable industry? Can they name the specific challenges they hit and how they solved them? References and case studies beat demos every time.
And lastly — and this one’s personal — use the tools you’re selling. I use AI every day in my own work. If you’re telling clients that AI will transform their business and you’re not using it to transform your own, they’ll sense the gap.
20. What kind of involvement do you maintain beyond your primary work?
I’m actively involved in the AI voice ecosystem beyond client work. I attend a lot of events and meetups in the
Bay Area, join virtual webinars, and make a point to stay on top of every meaningful development in AI voice. This space moves fast, and the only way to keep up is to be in the rooms where practitioners are sharing what’s actually working — not just what looks good in a press release.
I spend a meaningful amount of time in conversations with operators — call center directors, VP of operations types, CX leaders — who are trying to figure out what’s real and what’s hype in AI. Not sales conversations. Just honest exchanges about what’s working, what’s failing, and what’s next. Those conversations inform our work more than any conference or analyst report.
I’m also focused on content and education — building resources that help non-technical executives understand AI implementation without the jargon. The goal isn’t thought leadership for its own sake. It’s making sure the executives who need this technology can evaluate it honestly and deploy it successfully, whether they work with us or not.