How We Win — The Fork

DevenMnA = DevenLite (Targets + Tasking) + standalone minimal DevenDiligence + standalone minimal integration-reliant VDR — one greenfield platform, no legacy drag.

That definition is Diego's own, verbatim from the scoping prompt. The product shape around it — the component blurbs that follow — comes from AI-assisted analysis that has not yet been verified against our actual domain model; treat the specifics as directional, and see §6.

Components re-homed from the family

• Targets + Tasking (the DevenLite scope) — low-friction target tracking and deal tasking.
• DevenDiligence, minimal — checklists plus a hosted general LLM for summarization; a shared-folder feel rather than an enterprise evidence vault.
• The data room, minimal — integration-reliant VDR: storage and permissions through partners, not built in-house.
• Quick-valuation micro-products (the SDE calculator line) — fast, opinionated deal math for smaller deals.
• A reporting floor — self-serve from day one, not a services overlay.

Everything generic — e-signature, escrow, wire mechanics — is integrated, never built. The posture is opinionated and less configurable than Devensoft by design: fewer choices, faster paths.

One architecture picture

A domain-driven design exercise (AI-generated seed, illustrative — not yet validated against our real domain) decomposes the platform into five bounded contexts:

Target Sourcing · Valuation · Due Diligence · Deal Execution · PMI

with the core sub-domain — the thing DevenMnA exists to be great at — the Synergy Modeling & Value-Realization Engine: modeling the value a deal promises, then tracking whether the integration actually realizes it. Target screening and diligence workstream management support it; the data room, e-sign and escrow are generic buys.

The contexts wire together deliberately: findings flow from Due Diligence into Valuation through a translation layer (a diligence finding becomes a dollar adjustment, not raw-log noise); risks flow into deal terms; closing a deal fires the event that bootstraps the integration plan; and integration results feed back into the next valuation. Four aggregates anchor it — the valuation model (synergies can never exceed their cap; price = baseline + synergies − liabilities), the diligence audit (nothing clears with an unmitigated critical finding), the deal contract (nothing closes with open conditions), and the integration plan (no "realized" milestone without finance-verified tasks). A nineteen-event deal lifecycle runs through all five contexts, and it maps cleanly onto the nine-stage lifecycle we use everywhere else in this plan.

The point of the picture is not the boxes. It is where the center of gravity sits: value realization — the back half of the deal — is the core. The next section is why that placement is the whole argument.

Every current column — the platform as it ships today, the locked plan's AI phases, even the DevenLite carve-out — is weakest exactly where DevenMnA puts its core.

Three facts, then the honest framing.

Fact one — the current platform's strength is real but shaped by twenty years of accretion

Its integration core — planning, execution, the dependency engine, the IMO dashboards — is genuinely strong, and our code-level review confirms it. But it is configuration-shaped strength: it works the way the platform works, for the aspects the platform models, and reaching anything it did not anticipate means configuring around a structure that was never designed for it. The front of the lifecycle is tracking, not sourcing. The middle — valuation, deal execution, regulatory workflow — is thin to absent. And the platform carries a fundamentals backlog (auth, reliability, deployment health) that the locked plan must spend real weeks on before any new capability lands.

Fact two — the locked plan's AI phases deepen that shape; they do not change it

Scored on the agreed lifecycle matrix, the plan's AI work through Phase 3 strengthens diligence, pipeline and integration where the platform already works — and leaves opportunity identification, deal execution, Day-1 cutover and regulatory workflow as thin as today. It builds zero natively-fit lifecycle stages. That is not a criticism of the AI plan — it was never scoped to be a lifecycle redesign — but it caps how far evolution can reach.

Fact three — the industry literature says the money is made or lost in the back half

Generic but consistent: 70–90% of M&A deals fail to deliver their expected financial value, and the root cause is almost always mismanaged post-merger integration, cultural friction, or poor leadership. A platform whose core is the value-realization engine is aimed at the exact phase where deals fail.

So the greenfield argument is placement, not novelty. A clean build lets the core sit at value realization without dragging twenty years of structure along: no config-first assumptions, no fundamentals tax before feature one, no enterprise-complexity tax in the middle stages, and generic capabilities (data room, e-sign, escrow) procured rather than grown.

On the agreed lifecycle-fit matrix, DevenMnA is the only column with natively-fit stages — four of nine — and the only column with zero absent stages.

Coverage strips (native / config / partial / marginal / absent)

How to read it honestly

• Due Diligence at 93 is the matrix's top score — and it is deliberate: DevenMnA's diligence scope is where we plan to invest the most. Rationale on record: "a software that doesn't aim to do well at this phase isn't going to do good enough in the other stages."
• Value realization at 87 sits above both today's platform (62) and the locked plan's endpoint (65) — the core-domain bet, scored.
• Regulatory at 40 is the only credit any column earns beyond marginal — scope call on record: "we likely must include some of this functionality" — but it stays partial. Regulatory workload is not a native stage even here.
• Close & Cutover at 75 rides a spec, not a build: "configurable checklists that require evidentiary attachments." Promising shape; the least-proven of the strong cells.
• Neither v23 nor v24 has a single native stage. Today's platform is strong at integration execution — but configuration-driven, and only for some aspects of the stage. The locked plan's AI work shifts partial stages to configurability; it builds no native stage.
• DevenLite's six absent stages are by design — it was scoped front-only. Inside DevenMnA that front-only scope becomes the front of a full-lifecycle product; standalone, it was never the answer.

Either path runs on the same core crew — Diego, Roman, and the contract engineering bench — and the same runway bought by a ~$1.9M revenue base. Here is what runs and what pauses under each.

What each path does

The Hedge column is a pointer, not a re-presentation: the full locked plan, its sequencing and its recommendation cards live in the main plan's recommendations matrix and roadmap module — this brief does not restate them.

The money and the exposure

• ~$1.9M projected FY26 revenue — all of it tied to the current platform. Every renewal in that book is a Devensoft renewal. (Internal projection, unaudited, point-in-time — provisional at every cite.)
• 82.7% closed-renewal retention is the honest baseline the bold path must survive under the hand-off maintenance contract: no fundamentals remediation, no AI phases, custom report development curtailed.
• $128K attrition (7 accounts) already in the FY26 book, with the largest direct renewals ($200K medical-device renewal; $350K open at the largest direct account) landing inside this window regardless of which path is chosen.
• The vendor cost delta is modest either way. Hedge: 3 Talmatic at $15k/mo (net +$5k/mo over the current ~$10k). Bold: KAZ hand-off retained (~$10k/mo) plus 2 Talmatic engineers ($10k/mo) — net +$10k/mo. Neither arm changes the $1.9M revenue arithmetic.

Named risks (no hedging language)

What we do not know, stated plainly. None of it is disqualifying; all of it is decision-relevant.

Each of these has a path to resolution (sizing deep-dive, verification pass, explicit amendment). None of them resolves itself — they resolve because leadership funds the next step. That is §7.

The decision: choose the bold path (DevenMnA) or the Hedge path (the locked plan). The unchosen path pauses — that is what makes it a fork.

What Eric and Thom decide

What would change the recommendation

• A sizing deep-dive that breaks the feasibility story. If a calibrated dev-week estimate shows the bold crew — two architects + two engineers at agentic leverage — cannot reach a revenue-capable DevenMnA inside the runway the ~$1.9M base buys, the bold path dies on evidence, not preference. This is the first gate and should precede any irreversible commitment.
• Verification that materially moves the matrix. The pending verification pass owns the AI-derived specifics and the one derived cell. If DevenMnA's advantage does not survive verification, the lifecycle-fit argument weakens to a tie with the Hedge plan — and ties go to the incumbent path.
• A retention shock. If the renewal book falters during the decision window, the runway arithmetic changes first and hardest against bold.
• A demand signal. Any concrete buyer evidence for a lifecycle-complete smaller-deal platform upgrades the bold case from placement-argument to market-argument.

If neither is chosen today

Deferring is a decision with a cost: the senior crew stays split, the fundamentals backlog ages, and the renewal clock runs. The fork's structure is built to be decidable now on the evidence we hold — with the sizing gate as the safety catch on the boldest version of it.

Restated as one line: pick the platform you want us to be in three years — the one we already own, evolved — or the one the lifecycle says a winner should be, built. The evidence for both is on the table; the unknowns are named; the next move is yours.