Bank Connectivity: Strategic Decision Support¶
Prepared for leadership discussion — March 2026
The Core Tension¶
Palm's positioning is "AI-native cash forecasting that sits on top of existing connectivity." Our ICP (ICP 3: Modern TMS Augmenters) assumes customers already have bank connectivity through a TMS like Kyriba or their ERP.
Reality from the pipeline tells a different story. The majority of prospects either lack existing connectivity, have fragile connectivity, or expect Palm to handle it end-to-end. The "sits on top of" narrative is being heard as "we can ingest any data you have" — and we're saying yes.
This gap between positioning and operational reality is now a strategic risk.
What Engineering Is Seeing¶
From Rodel (VP Engineering):
- Current operational load: ~1 FTE absorbed by bank connectivity issues, trending toward 2-3 FTE
- Existing customers without connectivity (Discogs, Personio) are generating ongoing data quality issues and custom integrations (e.g., bespoke BlackRock ingestion for Personio)
- New prospects (Digital Realty, Euroports, Dunelm) are pushing for direct connectivity, compounding the load
- No dedicated investment has been made in the data ingestion stack to handle this at scale
- Engineering attrition and burnout risk is real if this continues unchecked
Key quote: "We did not invest time in our data ingestion stack to be able to do this for multiple prospects without drowning in operational load. This also is a hard pre-requisite for us being able to build a useable AI layer."
What the Pipeline Shows¶
An analysis of all sales prospects (Jan 30 – Mar 4, 2026) reveals the connectivity demand across ~25 active prospects:
Connectivity reality across the pipeline¶
| Category | Count | Examples |
|---|---|---|
| No existing connectivity — expect Palm to handle it | ~8 | Euroports, Highland Partners, Superdry, CRCE, Swiss Steel |
| Broken/limited connectivity — TMS has issues | ~5 | Bio-Rad (Kyriba API issues), Digital Realty (bypassing Cariba), Dr. Max (Nomencia overlay) |
| Mature connectivity — true "sit on top of" | ~3 | SIG Group (Kyriba 20yr), DAZN (Oracle Fusion APIs) |
| Too early / paused / budget-constrained | ~9 | Genesco, FG Europe, Omnion, Yondr |
Only ~3 out of ~25 prospects match our ICP assumption of having robust existing connectivity.
Connector demands surfaced in pipeline¶
| Type | Connectors Needed | Prospects Affected |
|---|---|---|
| European banks | ABN Amro, KBC, BNP Paribas, Greek banks | Euroports, Nash Squared, Brevo, Saracakis |
| US banks | Ameris, Bank of America, PNC | Highland Partners, Genesco |
| TMS integrations | Kyriba (robust), Nomencia, Cobase, Crown Data | SIG, Bio-Rad, Dr. Max, TBAuctions, TI Auto |
| ERP integrations | IFS, Oracle Fusion, NetSuite, SAP, D365, legacy (Coder, Unit Four) | CRCE, DAZN, ThinkProject, multiple others |
| Specialized | SEHA (Turkish), invoice financing platforms | BRF, Nash Squared |
Immediate revenue at risk from connectivity gaps¶
| Prospect | Est. ACV | Gap | Timeline |
|---|---|---|---|
| Euroports | ~EUR 60-80K | ABN Amro connector | Decision imminent |
| Highland Partners | ~$50K | Ameris Bank connector | March decision |
| Nash Squared | ~EUR 50-80K | KBC invoice financing | H2 2026 |
| SIG Group | ~EUR 50-80K | Deep Kyriba integration | April demo |
| Bio-Rad | TBD | Kyriba API issues | March PoC |
Estimated immediate pipeline at risk: $175-230K in ACV
The Strategic Question¶
Should Palm become a connectivity provider, or should we double down on being an intelligence layer that requires existing connectivity?
This isn't binary — but the answer shapes ICP, sales motion, engineering investment, and product roadmap for the next 12+ months.
Three Strategic Options¶
Option A: Stay the Course — "Intelligence Layer Only"¶
Reaffirm that Palm sits on top of existing connectivity. Tighten ICP to require prospects with working TMS/ERP connectivity. Accept narrower funnel.
| ICP impact | Narrows to ICP 3 (TMS Augmenters) with verified connectivity. Likely excludes ICP 2 (Spreadsheet Treasurers) entirely. |
| Sales impact | Smaller addressable pipeline. Longer qualification. Cleaner deals. |
| Engineering impact | Free up 1-3 FTE from connectivity ops. Focus fully on AI/forecasting/scenarios. |
| Revenue impact | Lose most of the immediate pipeline at risk. Bet on higher-quality, faster-closing ICP 3 deals. |
| Risk | Market may be too small if most mid-market treasuries lack good connectivity. Could limit growth when Palm needs logos. |
Option B: Build Connectivity as a Platform Capability¶
Invest in a scalable data ingestion and connectivity layer. Make bank/ERP connectivity a first-class product capability — not ad-hoc engineering work.
| ICP impact | Expands to ICP 2 + ICP 3. Opens up the "no TMS but needs one" segment. |
| Sales impact | Broader funnel. Connectivity becomes a selling point, not a risk. |
| Engineering impact | Significant upfront investment (dedicated team, connector framework, ops tooling). Reduces ad-hoc load over time. |
| Revenue impact | Captures the $175-230K at risk. Unlocks new segments. |
| Risk | Competes with established connectivity players (TIS, Cobase). Spreads engineering thin. Could delay AI/forecasting roadmap (scenarios, investments, explainability). |
What "invest properly" might look like:
- Connector framework with standardized adapter pattern (MT940, CAMT, BAI2, CSV, API)
- Dedicated connectivity team (2-3 engineers) or outsourced connector development
- Self-service onboarding for standard formats
- Monitoring/alerting for data freshness and quality
- Clear SLAs per connector type (standard vs. custom)
Option C: Selective Connectivity — "Strategic Bridges"¶
Build connectivity only where it unlocks high-value segments or is required by the ICP sweet spot. Establish clear criteria for when Palm handles connectivity vs. when the customer must bring their own.
| ICP impact | ICP 3 primary, ICP 2 selectively (if standard connectivity suffices). |
| Sales impact | Requires a "connectivity readiness" qualification step. Some deals get a "we'll build it" commitment, others get declined or deferred. |
| Engineering impact | Moderate investment in a connector framework + 2-3 high-value connectors. Reduces ad-hoc work without going all-in. |
| Revenue impact | Captures highest-value at-risk deals. Declines low-ROI custom work. |
| Risk | Requires discipline to say no. "Selective" can easily slide into "everything" under sales pressure. |
Possible selection criteria:
- Build if: connector serves 3+ prospects/customers, standard format (MT940/CAMT/BAI2), or unlocks a strategic logo
- Partner/refer if: niche bank, complex regulatory environment, or one-off ERP
- Decline if: legacy system with no API, single prospect, low ACV
Cross-Cutting Decisions (Required Regardless of Option)¶
1. Immediate messaging correction¶
The "we connect to everything" narrative needs to stop today. Replace with honest, tiered messaging:
- "Confirmed working" — MT940/CAMT processing, SFTP/email ingestion, Kyriba API (basic)
- "Available on request" — specific connectors Palm has built or can build in a defined timeline
- "Customer-provided" — connectivity Palm does not handle; customer brings their own data feed
2. Sales qualification gate¶
Add a mandatory connectivity assessment before any commitment:
- What data sources exist today?
- What format? (MT940, API, CSV, manual)
- Who manages connectivity currently?
- Is there a working TMS/ERP integration?
Engineering must sign off before sales commits to connectivity scope.
3. Operational triage for current customers¶
The existing operational load (Discogs, Personio, Digital Realty) needs a plan regardless of strategic direction:
- Stabilize current integrations (reduce firefighting)
- Document what exists vs. what's duct-taped
- Identify which current setups can be templated vs. which are truly bespoke
Framing for Discussion¶
| Question | Why It Matters |
|---|---|
| How many of our next 10 customers will have existing connectivity? | If the answer is "few," Option A may not be viable for growth. |
| What's the true cost of ad-hoc connectivity today? | 1 FTE? 2? Include opportunity cost of delayed roadmap items. |
| Can we afford to delay scenarios/investments/AI for connectivity investment? | These are the features that differentiate Palm from TMS players. |
| Is there a partner play? | Could Palm partner with a connectivity aggregator (TIS, Cobase, Plaid-for-treasury) instead of building? |
| What's the minimum viable connectivity investment? | Even Option A needs some investment to stabilize what we have. |
| Where does AI fit in? | Rodel flags that a reliable data layer is a prerequisite for the AI layer. If data is unreliable, AI features built on top will underperform. |
Recommendation Sketch¶
This section is intentionally left as a starting point, not a conclusion — the decision belongs to the group.
The status quo (doing nothing) is not viable. Engineering is absorbing unplanned connectivity work that will scale linearly with new customers. This crowds out roadmap ambitions and creates burnout risk.
The most likely sweet spot is Option C (Selective Connectivity) with clear guardrails:
- Invest in a connector framework that makes standard formats (MT940, CAMT, BAI2) self-service or near-self-service
- Build 2-3 strategic connectors (Kyriba depth, one major EU bank, one US bank) based on pipeline concentration
- Establish a firm qualification gate so sales never commits to connectivity without engineering sign-off
- Explore partnership with connectivity aggregators for the long tail of banks/ERPs
- Protect the AI/forecasting roadmap — connectivity investment should enable it, not replace it
The key insight from Rodel: connectivity and AI are not competing priorities — connectivity is a prerequisite for AI. The question is how much of that prerequisite Palm builds vs. requires customers to bring.
This document is for internal discussion. Data sourced from sales prospect analysis (Jan 30 – Mar 4, 2026), VP Engineering input, domain knowledge base, and ICP strategy docs.