Interactive Demo

Deep Research + Hermes

Three pre-loaded prospects. Walk through the agent output. Built by Joseph DeVille.
How this works: Pick a prospect below. The page shows the research packet Deep Research pulled, the persona read it inferred, the close plan it recommended, and three Hermes-generated outbound variants. Click any variant to see the learning signal that would re-rank it for future sends to this persona. Start with prospect 2 for a regulated, consensus-buying motion similar to fleet compliance sales.

Jamie K. — VP of Engineering

at Fictive — a 450-person developer tools SaaS, Series C
Dev Tools Technical, Decisive
Research Packet

Company signals

  • Raised $60M Series C, March 2026
  • Posted 12 platform-eng job reqs in last 90 days
  • Stack: GitHub, AWS, Datadog, PagerDuty, Backstage (self-hosted, 2+ years)
  • Engineering blog: 3 posts in Q3 on "scaling Backstage"

Attendee profile

  • VP Eng for 14 months, prior Director at Stripe
  • CS degree, 11 years in infra / platform
  • Active on LinkedIn: posts about developer productivity
  • No prior touches from us
Persona Read
Data-driven Decisive Build-vs-buy skeptic Owns the budget Will say no in 30 seconds
Technical-depth buyer. Will not sit through a demo that doesn't open with architecture. Trusts peer-reviewed numbers over vendor claims. Already publicly committed to a Backstage build, so the pitch is not "switch tools" — it's "stop paying the tax of maintaining one."
Close Plan
  1. Open with the Backstage tax. Lead with time spent on plugin maintenance per quarter at companies their size (~400hrs eng).
  2. Lead product angle: managed scorecards + ownership, not "replace Backstage." Co-exist first, consolidate later.
  3. Likely objection: "we already invested in Backstage." Pre-empt with 2 case studies of Series C dev-tools companies that ran both for 90 days and killed Backstage after.
  4. Stall point: security review. Have SOC 2 + the self-hosted deployment doc ready for the second call.
Hermes Outbound Variants
Variant A
The Backstage math at your scale
Jamie, Saw the Q3 posts on scaling your Backstage build. At ~450 engineers, the average we track for Backstage plugin maintenance is 380-420 eng hours per quarter. Three Series C dev-tools companies ran Cortex alongside Backstage for 90 days this year. All three killed the self-hosted instance after. Worth 20 minutes to see the architecture side-by-side? Joe
Hook: cost, specific hours. CTA: 20 min, architecture frame.
Variant B
Question on your Backstage rollout
Jamie, Short question. You've been shipping Backstage improvements publicly for 2 years. At what point does the plugin maintenance stop being a feature of your platform team and start being a tax? Asking because we see the inflection usually hit between 400 and 600 engineers. You're inside that window. Open to compare notes? Joe
Hook: question, respects the prior investment. CTA: peer frame.
Variant C
Fictive is hiring 12 platform engineers
Jamie, 12 platform eng reqs open since August. The usual reason at this stage: the internal tools that got you to 400 engineers need to get you to 1,000, and the team to maintain them is the hiring bottleneck. Cortex lets a platform team of 6 do the work of 12. Happy to show how, 20 min. Joe
Hook: hiring signal. CTA: ratio-of-people frame.
Learning signal:

Patricia M. — Head of Operations

at Cole & Reeves — regional law firm, 180 attorneys across 4 offices
Legal Services Consensus, Risk-aware
Research Packet

Firm signals

  • Opened a 4th office (Nashville) in June 2026
  • Practice mix: corporate, IP, employment, litigation
  • Stack: iManage, NetDocuments, Clio Manage, Outlook
  • Recent press: internal AI task force launched Q2

Attendee profile

  • Head of Ops for 6 years, prior firm admin at an AmLaw 200
  • No technical background, buys through committee
  • LinkedIn: active in legal ops community, posts on ILTA
  • Prior touch: downloaded a white paper on AI governance in May
Persona Read
Consensus buyer Risk-first Needs legal + IT buy-in Reads every clause Slow yes, hard no
Operates in a profession where being wrong is expensive. Will not move without IT security sign-off, managing-partner buy-in, and at least one peer firm reference. Anything that reads as "move fast and break things" dies on arrival. Frame everything in terms of controls, auditability, and reversibility.
Close Plan
  1. Open with governance, not productivity. Lead with how the solution handles data residency, privilege, and audit trail. Features come after.
  2. Lead product angle: matter-level controls and attorney-review gates. Not autonomy, oversight.
  3. Likely objection: "we'd need IT to evaluate." Pre-empt with a SOC 2 Type II report + the deployment doc in the first email.
  4. Reference strategy: one peer firm (similar size, similar practice mix) that will take a reference call. Named firm matters more than number of logos.
  5. Buying committee: Managing Partner (final yes), IT Director (security veto), 1-2 practice-group chairs (adoption).
Hermes Outbound Variants
Variant A
AI governance framework for mid-market firms
Patricia, Saw Cole & Reeves stood up an AI task force earlier this year. At 4 offices and 180 attorneys, the governance piece is usually what stalls deployment. We put together a framework used by 3 firms similar to yours (data residency, privilege controls, matter-level audit trail). Happy to send the one-pager and compare notes on what your task force has landed on so far. Joseph
Hook: governance. CTA: send doc, no meeting ask.
Variant B
Nashville office — ops question
Patricia, Congrats on the Nashville launch. A pattern we see with firms opening a 4th office: the internal tools that supported 3 offices quietly stop scaling, and ops absorbs the gap. Would it help to compare what firms your size are doing to avoid that, specifically on matter intake and conflicts across offices? No pitch, 15 min call. Joseph
Hook: expansion moment, ops pain. CTA: 15 min, no-pitch frame.
Variant C
Reference call with Henderson Prescott?
Patricia, Short note. Henderson Prescott (210 attorneys, similar practice mix to Cole & Reeves) agreed to take reference calls from firms considering the kind of AI governance work we do. If it's helpful, I can broker the intro. They went live in March and could share what worked and what didn't on the IT security review. Joseph
Hook: peer reference first. No product mention.
Learning signal:

Daniel R. — Director of Finance

at Vessel — Series B payments infrastructure, 95 employees
Payments Analytical, Owns Numbers
Research Packet

Company signals

  • Series B $28M, April 2026
  • Vertical: SMB card-present payments, LATAM expansion
  • Stack: Netsuite, Ramp, Stripe (internal billing)
  • Hiring 2 FP&A analysts, posted Oct

Attendee profile

  • Director of Finance, 2 years at Vessel, prior Sr Mgr at Plaid
  • MBA, CFA level II, built FP&A function from scratch
  • Active on Twitter (fintech twitter), replies in finance ops threads
  • Prior touch: attended a webinar on Series B unit economics, July
Persona Read
Owns the model Numbers-first Suspicious of vendor claims Will test you on math Reports to CFO + board
Will open the first email with the spreadsheet mindset on. Loses respect fast if a vendor misstates a benchmark or quotes a figure without a source. Needs the numbers to defend internally more than they need the product to work — because they're the one who has to defend them.
Close Plan
  1. Open with a specific number, cited. Not "save you time." Give him the real benchmark with the source.
  2. Lead product angle: close-the-books cycle time, with a before/after at a comparable payments company.
  3. Likely objection: "we're already building this internally with the new FP&A hires." Pre-empt with the build-vs-buy math at Series B headcount.
  4. Buying committee: Director Finance (champion), CFO (approver), Head of Eng (integration veto).
Hermes Outbound Variants
Variant A
Close-the-books benchmark for payments at your stage
Daniel, Benchmark we pulled for Series B payments companies (n=18, headcount 60-120): median monthly close is 11 business days. Top quartile is 6. Vessel's at 95 employees, LATAM expansion live. The LATAM entity work is usually where the quartile gap opens up. Happy to send the full dataset and walk through what the top quartile is doing differently. Joe
Hook: cited benchmark, specific to stage. CTA: dataset, numbers-first.
Variant B
The FP&A hire math
Daniel, Saw the 2 FP&A reqs open. Standard build-vs-buy question at Series B: is it faster to hire 2 analysts and write the automation internally, or buy the tool and have the same 2 analysts doing higher-leverage work? At 95 employees the math usually favors buy by ~7 months of payback. Happy to send the model. Joe
Hook: hiring signal, build-vs-buy math.
Variant C
Plaid -> Vessel transition — close velocity
Daniel, You moved from Plaid to Vessel 2 years ago. Common pattern for finance leaders making that jump: the tooling at Plaid was mature, the tooling at the next company isn't yet, and the first 18 months are spent building what you took for granted. If Vessel's still in that build-out phase, there's a shortcut worth 20 minutes. Joe
Hook: career-arc pattern, respect-the-journey frame.
Learning signal: