Product Vision / v2 · Board-Approved

Ship the Positioning. Prove the Product.

"I couldn't take that job on because these types of personas are too expensive to get a hold of. And even if I could, I could never get them into a room together to debate." Researcher, Australia

The positioning shift is free and ships this week. The research foundation is not a 12-month roadmap, it is a funded pilot. One vertical. 10-15 interviews. One blind A/B quality test. Then decide. The board consensus: compelling vision, premature commitment.

The Insight Behind the Insight

It Only Works If It Feels Real

A synthetic panel of "generic CHROs" is a parlour trick. A synthetic panel of CHROs trained on 50 depth interviews with real CHROs across fintech, healthcare, and enterprise SaaS is a research instrument. The difference is the training data, not the technology.

The question is not "can AI simulate a CISO?" The question is "have we done the foundational research to know what a CISO actually thinks, worries about, and optimises for?" If yes, the synthetic panel is credible. If no, it is fiction.

The Limit of Current Synthetic Qual

Today's evaluation gates catch methodological failures: leading questions, confirmation bias, unrealistic consensus. They do not catch shallow personas. A panel can pass every quality gate and still produce insights that feel generic, because the personas were built from public knowledge rather than proprietary research.

The fix is not better prompts. It is better inputs. Real interviews. Real data. Real depth.

The Value Chain

1
Deep Interviews
50+ depth interviews with real professionals in a specific vertical and role. "We talked to 50 revenue leaders in fintech." This is primary research, conducted once, used many times.
2
Calibrated Personas
Interview data trains calibrated synthetic personas that reflect real decision-making patterns, real objections, real language. Not "what would a generic CISO say" but "what would a CISO who has lived through a breach, managed a $5M security budget, and reports to a sceptical board say."
3
Reusable Expert Panels
The calibrated personas become a standing panel. Multiple clients can run different questions against the same research-backed panel. The panel gets better with each study as edge cases surface and the training data expands.
4
Ongoing Access
Clients subscribe to panel access. New questions, new scenarios, new concepts tested against the same calibrated experts. The panel is a living research instrument, not a one-off study.
Why This Is a Different Business (If Proven)

Today: project revenue. Client pays $1,950-4,500 per study. Relationship is transactional.

Tomorrow, if the pilot works: platform revenue. Client pays for ongoing access to a research-backed expert panel trained on real interviews. Relationship is recurring. The more studies run against the panel, the more valuable the panel becomes.

Sparring Partner's Warning

Rob is a solo founder. The platform vision is correct but the full investment is premature. The pattern to resist: big vision, big build, no first-step validation. A $10-15k per-vertical sprint commitment before a single client has paid for research-backed output is the exact mistake that kills bootstrapped businesses.

The discipline: positioning ships now because it costs nothing. The research sprint is gated behind a pilot that proves clients can feel the quality difference. No full verticals until there is revenue to fund them.

Product Architecture

The Research Foundation

Each vertical panel starts with a research sprint: 50+ depth interviews with real professionals in the target role. These interviews follow a structured protocol designed to capture decision-making patterns, not just opinions. What triggers their decisions. What they optimise for. What they fear. What they have seen fail.

This is traditional qualitative research done excellently, once, to power synthetic research at scale.

Three Panel Types

Vertical Panels (Built by Pythia)

Pre-built panels for high-demand verticals: fintech leadership, cybersecurity decision-makers, healthcare C-suite, enterprise HR. Trained on Pythia's own research sprints. Available to any client. The "off the shelf" product.

These are the panels that demonstrate credibility. "Our fintech leadership panel is trained on interviews with 50 revenue leaders across seed-stage to Series C." That is a proof point traditional research cannot match, because no recruiter has done it either.

Bespoke Panels (Commissioned by Client)

Client commissions a research sprint for their specific domain. "We need a calibrated panel of Australian mining safety officers" or "We need enterprise procurement leaders in DACH." Pythia runs the interviews, builds the panel, and the client gets exclusive or semi-exclusive access.

This is the service business. Higher price point. Deeper relationship. The client's investment in the panel makes switching costs real.

Hybrid Panels (Cross-Functional Assembly)

The killer application: assembling cross-functional panels from different vertical libraries. A cybersecurity vendor needs CISO + CFO + Board perspectives. Pull calibrated personas from three different vertical panels, put them in a room. This is the "impossible assembly" that Craig identified. The research foundation makes it credible.

What Makes This Defensible

The Data Moat

The technology layer (running synthetic panels, moderating debates, producing reports) is replicable. The research foundation (50+ interviews per vertical, calibrated personas, validated against real outcomes) is not. Every research sprint Pythia runs deepens the moat. Every client engagement generates calibration data. Competitors would need to do the same interviews to match the quality.

50+
Interviews per vertical panel
8-12
Calibrated personas per panel
Reusable
Panel improves with each study
Cross-functional
Mix panels for impossible assembly

Where Research-Backed Panels Win

These are the verticals where the "impossible panel" problem is acute and the willingness to pay is highest. The research sprint investment is justified by the number of buyers who need the panel.

Tier 1: Build These First

Fintech Leadership Panel
CROs, CFOs, Chief Compliance Officers, VP Product at $50M-2B fintech companies
Impossibility
5
WTP
5
Buyers
1,500+

Research sprint: 50 interviews with fintech CROs and CFOs. Decision patterns around pricing architecture, compliance trade-offs, board dynamics. Use cases: Pricing model validation. Go-to-market pivots. Regulatory strategy. Competitive positioning.

Why this first: Highest composite score. Warm channel through VC portfolio companies. Multiple repeat-use scenarios per client.

Cybersecurity Decision-Maker Panel
CISOs, Board Audit Chairs, CFOs, VP Security at enterprise
Impossibility
5
WTP
4
Buyers
4,500+

Research sprint: 50 interviews with CISOs and security leaders. How they evaluate vendors. What gets board approval. What kills deals. Use cases: Vendor positioning. Board objection mapping. Competitive differentiation. Sales enablement.

Why this second: Largest addressable buyer pool. Clear pain point (CISO says yes, board says no). Cybersecurity vendors have budget and urgency.

Tier 2: Build Next

Healthcare C-Suite Panel
CMOs, Hospital CFOs, Chief Nursing Officers, Patient Safety Officers

Research sprint: 50 interviews with health system leaders. Care delivery innovation, staffing models, technology adoption. Buyers: ~1,500 health systems + healthtech startups.

Pharma/Medtech Regulatory Panel
Chief Regulatory Officers, VP Medical Affairs, General Counsel

Research sprint: 50 interviews with regulatory leaders. Submission strategy, accelerated pathways, compliance architecture. Buyers: ~350 FDA-regulated companies + investors. Highest WTP per buyer.

Enterprise HR / People Leadership Panel
CHROs, Chief Talent Officers, VP Compensation

Research sprint: 50 interviews with HR leaders. Total rewards, salary transparency, remote work policy, talent strategy. Buyers: ~600 enterprises + HR tech companies. This is Craig's original use case.

Tier 3: Market Signals

Vertical PanelResearch SprintEst. Buyers
AI/ML Infrastructure Leaders50 VP Eng / CTO interviews~800
Defence / Gov Compliance50 contracting officers~1,500
Sustainability / ESG Leaders50 CSO / ESG interviews~1,200
VC / PE Investment Partners50 GP / LP interviews~250
Data Governance Leaders50 CDO / CPO interviews~1,400

The Economics

Revenue Architecture

Tier 1
Decision Pack
$1,950
Single study against a pre-built vertical panel. 2-3 concepts tested, 48hr delivery. The entry point.
Tier 2
Panel Subscription
$4,000/mo
Ongoing access to a vertical panel. 3 Decision Packs per month. Panel improves with each use. The retention play.
Tier 3
Bespoke Panel Build
$15-25k
Commissioned research sprint: 50+ interviews, calibrated persona set, exclusive panel access. The premium service.
The Research Sprint as Investment, Not Cost

Each vertical panel requires ~50 depth interviews. At $200-300 per interview (industry standard for executive recruitment + incentive), the research sprint costs $10-15k in direct costs plus research time.

If the resulting panel serves 20 clients at $1,950 per study, the sprint generates $39k in first-year revenue against $10-15k investment. If 5 of those convert to $4,000/mo subscriptions, that is $240k ARR from a single research sprint.

The economics improve with each subsequent vertical because the methodology, interview protocol, and calibration process are reusable. Sprint 2 is cheaper and faster than Sprint 1.

Traditional Qual vs. Research-Backed Synthetic

DimensionTraditional QualResearch-Backed Synthetic
Panel recruitment$15-60k per study$10-15k once, reused across clients
Timeline per study4-12 weeks48 hours
Cross-functional assemblyOften impossibleMix panels from vertical libraries
Panel quality over timeStarts fresh each timeImproves with each study
Client cost per study$15-60k$1,950-4,500
Fraud risk54-88% on exec panelsZero (trained on verified interviews)
Honesty constraintSocial/institutional pressureNone (no career risk for personas)
The Researcher Angle

Researchers who currently turn down work because recruitment is impossible become the people who conduct the foundational interviews. Craig does not become a Pythia client. Craig becomes a Pythia research partner. He runs the depth interviews that train the panels. His expertise makes the synthetic output credible. His name on the research sprint is a proof point.

This turns a coaching relationship into a revenue partnership.

Persona-as-a-Service: The Endgame

If Pythia builds 10 vertical panels, each trained on 50+ interviews, that is a library of 500+ calibrated professional personas covering the hardest-to-reach segments in qualitative research. No company on earth has that. Not Kantar. Not Ipsos. Not McKinsey.

That library becomes a platform. Other research firms, consultancies, and agencies license access to run their own studies against pre-calibrated panels. Pythia becomes infrastructure for professional qualitative research, not a service provider.

What's Next

The Core Question

This vision is compelling. The question is sequencing. Building one research-backed vertical panel well proves the concept. Building ten poorly proves nothing. The first sprint needs to be excellent, visible, and commercially successful.

Phase 1: Prove It (Months 1-3)

1
Pick one vertical and run the research sprint. Recommendation: fintech leadership. Highest composite score, warm VC channel for both interview recruitment and client acquisition. 50 interviews with CROs, CFOs, and compliance leaders across seed-to-Series C.
2
Build the calibration protocol. How do real interview data become calibrated personas? This is the methodological IP. Sable owns this. It needs to be documented, repeatable, and auditable.
3
Run 3-5 pilot studies against the panel. Free or deeply discounted. Generate case studies that show the difference between generic synthetic and research-backed synthetic. The quality gap should be obvious.
4
Explore the Craig partnership. Can he run or co-run the research sprint? His credibility as a traditional researcher endorsing the approach is worth more than any marketing. "I used to turn these briefs down. Now I run the interviews that make them possible."

Phase 2: Commercialise (Months 3-6)

5
Launch the fintech panel commercially. Decision Pack pricing against the research-backed panel. Measure quality difference vs. current output. Publish the methodology (not the data).
6
Start the second vertical sprint. Cybersecurity, based on buyer pool size. Use learnings from Sprint 1 to reduce cost and timeline.
7
Introduce the Panel Subscription tier. $4,000/mo for ongoing access. Target independent strategists and repeat-use enterprise clients. This is the recurring revenue engine.

Phase 3: Platform (Months 6-12)

8
Bespoke Panel Build as a service. Clients commission research sprints for their specific domains. $15-25k per sprint. Higher margin, deeper relationships, real switching costs.
9
Cross-functional assembly. Mix personas from different vertical libraries. CISO from cyber panel + CFO from fintech panel + Board member from governance panel. The "impossible room" that Craig described, built from real research across multiple domains.
10
Evaluate licensing model. Other research firms and consultancies access pre-calibrated panels for their own studies. Pythia becomes infrastructure. This is the platform play, but only if the panel quality is demonstrably superior.
The Test for Every Decision

At each step, the question is: "Is the research-backed panel producing insights that are specific enough to feel real?" If a client reads the output and says "this sounds like it was written by someone who has actually done this job," we are on track. If they say "this sounds like ChatGPT," we have more work to do on the training data. The quality bar is not "passes our evaluation gate." The quality bar is "a practitioner would nod."

Pythia · Impossible Panels: Product Vision · v2.0 · April 2026