How to spec an engineering pod in 30 minutes: the operator’s checklist for a discovery call

Most “discovery calls” I sit on the other side of waste 20 of the 30 minutes on vendor pitch. The buyer joins with a roadmap PDF and a vague budget. The vendor opens with a six-slide deck, a logo wall, and a “tell me about your stack” question that the buyer already answered in the intake form. At minute 28 someone says “let me put together a proposal and send it over by Friday.” Friday becomes Tuesday. Tuesday becomes a redline cycle. Three weeks later the buyer is still un-podded, still behind on the roadmap, and now ramping a different vendor with the same script. The 30-minute call is a procurement artifact, not a scoping instrument. It does not have to be. This post is the operator checklist that fixes it, written from the chair I sit in at Devlyn AI, where 30-minute calls produce written pod specs roughly 7 times out of 10.

TL;DR

• A real 30-minute discovery call ends with a written pod spec, not a “we will follow up.”
• The buyer’s homework is heavier than the vendor’s. Bring 10 things, leave with a number, a composition, and a ramp date.
• The trial scope, the budget envelope, and the named stakeholder are the three items that decide whether minute 28 produces a contract or a stall.
• Devlyn discovery calls run against a 3-day free trial and a 14-day replacement guarantee. Both belong in the spec, not the small print.
• The buyer who walks in prepared converts the call at 70%. The buyer who improvises converts at 15%. The asymmetry is structural, not vendor-side.

What a 30-minute call should actually produce

A written pod spec. One page. Composition (e.g., two senior engineers, one PM, shared DevOps at 0.5 FTE), pricing (a monthly retainer number, not a range), ramp date (the actual Monday the pod shows up in your Slack), and the named risks the pod owns (architecture decisions, security review, on-call rotation, integration pipeline, whatever applies). The spec also names what the pod does not own, so the boundary is clear before contract.

Compare that to a typical agency call. Typical output: a follow-up email with a calendar link for “round two”, a sales engineer attached, and a proposal pencilled for the following week. That is two weeks of buyer time spent for a document that arrives Day 14. By Day 14 a real operator has already shipped the trial against a different vendor or pulled an internal stretch assignment. The 30-minute call is a buyer’s instrument. The buyer sets the standard. Devlyn calls produce the one-page spec because we built the call format around it; not every vendor does, which is the buyer’s filter.

The checklist, 10 items you bring to the call

1. The roadmap item this engagement closes. Name the deliverable and the date. “Migration off legacy Twilio to Vonage by August 15.” Not “improve communications infrastructure.” Vendors who can scope cannot scope vapor. The roadmap item is the anchor for every downstream decision: composition, trial ticket, success metric. If you cannot name one in 30 seconds, you are not ready for the call, you are ready for an internal alignment meeting.

2. The current team shape and what is blocking it. Five engineers, one tech lead, two on hire-and-onboard, one on PTO through July. The blocker is usually one of three: capacity, specialist skill, or velocity per engineer. Name which. “We have headcount, we lack a Postgres performance specialist for two months” is a different pod than “we need 4× the roadmap velocity at the same team count.” Devlyn’s pod compositions for Calenso, Creator.ai, and Klaviss each solved a different one of those three shapes. The vendor’s job is to read which yours is in minute 4.

3. The stack and compliance posture. Named technologies (Laravel 11, React 19, Postgres 16, AWS us-east-1, Snowflake), plus the compliance frame (HIPAA, SOC 2 Type II, GDPR, PCI-DSS, or “none yet, Series A”). The compliance posture decides whether the pod needs a security engineer in the composition. A Haxi.ai-style AI engagement at scale and a healthcare SaaS chasing HIPAA produce different pod specs even when the stack overlaps 80%.

4. The metro alignment and time-zone window. Where do your engineers sit and what hours do you expect overlap? “We are New York and Berlin, we want 4 hours of overlap in the Berlin afternoon” is enough. The vendor needs this to staff the pod from the right roster. Skip this and you get a pod that ramps clean and then misses standups for the first two weeks while everyone renegotiates.

5. The 90-day success metric. One number, one deadline. “32 PRs merged through main by Day 90” or “the Vonage migration shipped, on-call rotation handed over, zero P1 regressions through 30 days of soak.” Vague metrics (“improve velocity”) produce vague pods. Specific metrics produce specific compositions and let both sides agree on the trip-wire for the 14-day replacement clause.

6. The trial scope, a real backlog ticket. Not a take-home. A live ticket you would otherwise hand to your team this sprint. Devlyn runs a 3-day free trial against the actual ticket; the output is production-quality code with tests, not a slide. Bring the ticket to the call. The vendor scopes the trial in minute 22. If the vendor’s first response is “we do not do trials” the call ends at minute 23.

7. The retainer budget envelope, a number, not a range. “We have $11,000 a month approved through Q3.” Not “somewhere between $8K and $15K.” Vendors who hear a range price to the top of it. Vendors who hear a number price to the spec. Devlyn pods start at $2,500/month for a single engineer and scale to $20,000+ per month for senior-heavy compositions with embedded PMs; the envelope decides which composition you see in minute 25.

8. The contract redline expectations. MSA terms you cannot move on: IP assignment (work-for-hire, full assignment on payment), liability cap (typically 12 months of fees), replacement clause (Devlyn writes the 14-day replacement guarantee directly into the MSA, not the marketing site), termination notice (30 days standard, 60 if your finance team requires). Bring your standard MSA if you have one. Vendors who cannot redline in the room cannot ship in 24 hours either.

9. Existing vendor relationships to coordinate with. If you use Remote.com or Deel for EOR, name them: the pod-to-FTE conversion path runs through whichever you already pay. If you run security review through Vanta or Drata, name that too. The pod’s first week is smoother by 4 to 5 working days when these integrations are pre-named.

10. The internal stakeholder who must sign off. CFO, board observer, head of security, whoever holds the “no” vote on retainers north of $X. Either bring them to the call or get explicit pre-approval for the envelope in item 7. Most stalls between minute 30 and contract are not vendor friction; they are an internal stakeholder discovering the engagement on Day 8 and resetting the clock.

What the operator brings to the same call

The vendor side of the table owes you symmetry. From Devlyn’s chair, that means three candidate pod compositions, not one. Composition A is the obvious read of your brief. Composition B is leaner, one less engineer, with named tradeoffs (slower ramp, narrower stack coverage, lower retainer). Composition C is heavier and faster, with the cost articulated openly. The buyer picks. The vendor does not push the middle one because the middle one is the easiest sale.

The 14-day replacement guarantee is written into the MSA before signature, with the trigger condition spelled out: missed standup pattern, PR rejection rate above threshold, security review failure, your call. The 3-day free trial ticket is scoped live in the call, with the deliverable, the acceptance criteria, and the engineer’s name attached. Pricing is the single number that matches the envelope you brought. The ramp date is a specific Monday. The pod’s first PR target is a specific Friday. The numbers are defensible because we have run the same composition at Calenso, Creator.ai, Klaviss, and Haxi.ai often enough to know what slips and what does not.

Compositions, replacement clause, trial ticket, ramp date, retainer number, and the one-page spec. Six artifacts. 30 minutes.

What kills the 30-minute call

Three anti-patterns, in descending frequency.

The vague roadmap. Buyer says “we need engineering capacity” without a deliverable or a date. The vendor cannot scope. The call becomes a vendor pitch by default because there is nothing else to fill the time. Fix: bring item 1 from the checklist.

No budget envelope. Buyer refuses to share a number, hoping the vendor will quote low. The vendor quotes the composition the brief implies and the buyer balks because the number is 2× what they had imagined. Fix: bring item 7. Real envelopes produce real specs; coyness produces a second call.

No stakeholder. The buyer is technically the decision-maker but the CFO holds an unsaid veto on retainers over $10K. The 30-minute call produces a clean spec; the spec dies in finance on Day 8. Fix: bring item 10, or pre-clear the envelope.

Run the call

If your engineering capacity is the constraint and you are tired of the 14-day proposal cycle, the checklist above is the homework. Bring it to a Devlyn discovery call. I will bring the three compositions, the trial ticket scope, and the one-page spec. If the math is closer than that, run it yourself against the pod ROI calculator first. If the question is whether Devlyn is the right shape of vendor, Trust is the page that names what we own and what we do not. The pod path itself, what we deploy and how it ramps, lives at dedicated developers and engineering pods; if you are scoping a captive offshore centre rather than a retainer, GCC is the next stop.

FAQ

Should I sign an NDA before the discovery call? No. Both sides can speak in patterns and named archetypes for 30 minutes. The NDA goes in writing before the 3-day free trial, because the trial means handing over a repo access token. NDAs as a prerequisite to a scoping call signal procurement theater, not security maturity.

What if I am not ready to commit? Run the 3-day free trial anyway. It costs nothing, the output is real working code, and the worst case is you walk away with a feature shipped. Most Devlyn engagements convert in the week after the trial; about 15% convert 60 to 90 days later when the in-house plan stalls.

Can we run a trial without a retainer? Yes. The 3-day trial is unconditional. The 14-day replacement guarantee kicks in if you sign the retainer. If you do not sign, the trial output is yours, full IP transfer, no clawback.

What does pricing look like for a Series B versus a $200M IT shop? Series B retainer engagements typically start around $2,500/month per engineer and pod compositions run $8,000 to $15,000/month. A $200M IT shop running a multi-pod programme with embedded PMs, dedicated DevOps, and security review sits in the $30,000 to $80,000/month range. The envelope determines the composition, not the company size.

Who attends the call from Devlyn’s side? One person, usually me. Not a sales engineer plus an AE plus a delivery lead. The 30-minute call is operator-to-operator, and the spec we produce is the one I sign off on. If the engagement requires a specialist on the call (HIPAA work, complex DevOps, AI/ML scope), I bring them; otherwise the buyer is talking to one person who can both scope and commit.