The AI Deal Playbook — why we generate an action plan for every insolvency case

In the last five years a dozen insolvency data providers have emerged in DACH: Insolvenz-Radar, InsolvenzIndex, STP Insolvenz-Portal, AssetRadar, Deal One and others. They all do essentially the same thing: aggregate § 9 InsO announcements, enrich with commercial-register and Federal-Gazette data, and wrap a filter around it.

That is valuable — especially compared to manually scanning insolvenzbekanntmachungen.de, which has no filters. But it doesn't answer the one question that keeps every buyer up at night: what do I concretely do with this case?

The path from raw data to action is the real work. Four to eight hours per qualified case if you want to do it right: pull register extract, get annual report from Federal Gazette, check brand portfolio, analyse web presence, research administrator history, set up outreach, prepare first meeting. And you carry that cost per case.

The Deal Playbook is an AI-generated action plan we produce for every case on our platform, personalised to YOUR buyer profile. It combines two inputs — the case dossier (enriched register data, Federal Gazette, web research) and your buyer profile (budget, revenue range, sector preference, procedure preference, deal thesis).

The output is a structured playbook with eight components: (1) fit score 0-100 with a clear verdict, (2) reasoning in 2-4 sentences — why this case fits you or doesn't, (3) estimated asset value as an EUR range, (4) budget alignment (is your budget above, at, or below the realistic price range), (5) structural recommendation (asset deal, selective asset deal, share deal via insolvency plan, or pass), (6) concrete focus assets calibrated to YOUR budget, (7) three key questions for the first call with the administrator (case-specific, not generic), (8) three case-specific risk flags.

And then the piece that makes the difference: a ready-to-send first-outreach email — proper salutation, company name, case number, concrete assets you're interested in, your positioning statement. Copy, send.

Technically it's unspectacular and therefore robust: we build a structured prompt from case data + buyer profile, send it to a mid-tier language model (Claude Haiku / GPT-4o-mini) and parse the response as strict JSON. The answer is cached per (case, profile) pair, so every repeat call loads instantly without regenerating cost.

The intelligence sits in the prompt design, not in model magic. We instruct the model explicitly: first hard filters (if documented revenue is outside the target range, fit score caps at 30), then fit analysis, then structural recommendation, then application to concrete assets calibrated to the budget. It works because we deliver case data in structured form — commercial-register basics, Federal Gazette metrics, administrator details — not as prose. The model has nothing to invent.

Cost per playbook: about half a US cent. Latency: 4-8 seconds. Scales easily to hundreds of thousands of case × profile combinations.

Most „AI features” in SaaS are marketing cashmere draped over a chat widget that answers from marketing collateral. The Deal Playbook is something else: it makes a concrete decision („buy or pass”), it names concrete numbers (asset value range), it produces an artefact (the email) the buyer can use immediately. The line between „pretty AI decoration” and „real tool” is exactly this: does an artefact result that a human uses, or does prose result that evaporates?

In distressed M&A time is the single most important competitive factor. Not capital, not network, not expertise — time. The buyer with a solid offer to the administrator in week 2 wins against the buyer with a question list in week 4. The Deal Playbook makes you faster.

Take a fictional case: „Sample Textiles GmbH”, Fashion D2C, last revenue EUR 2.4m, 12 employees, in Eigenverwaltung for 6 weeks. Administrator: Dr. Maier from Wellensiek.

Buyer Profile A („consolidator”): family office, budget EUR 500k-2m, focus D2C fashion, retention-marketing expertise. Playbook: fit score 82 (strong fit). Structure: asset deal with staff transfer under § 613a. Focus assets: brand, customer database, inventory (below 20 % of market), 6-8 core employees. Email draft: positioning as consolidator, concrete reference to prior deal, meeting proposals.

Buyer Profile B („micro-deal operator”): 2-person team, budget max EUR 10k, seeks brand + customer list to continue in own setup. Playbook for the SAME case: fit score 34 (poor fit). Structure: pass — or alternatively: very selective asset deal (brand + domain only, if the administrator would accept it under a residual-monetisation angle). Focus assets narrowly scoped: brand, domain, order history, no staff, no inventory. Email draft: minimal positioning, direct signal „pure IP interest”.

One case, two playbooks, two entirely different action recommendations. That's the point: reality never fits on a single „match score” number.

The playbook is version 1.0. On the roadmap: (a) auto-generation for the top match of the day in the morning email — so every user starts the day with a case + playbook on the table, without searching. (b) Buyer Profile v2 with a structured form (budget, revenue range, sectors as explicit fields alongside prose). (c) Playbook chat — a conversation layer on top of the generated playbook for follow-ups and negotiation coaching.

If you want to see one live — sign in with a buyer profile, open any dossier on /unternehmen. In ten seconds you have the action plan.