10.4 Evidence Quality Addendum (AI-Native Automation)
This addendum defines evidence priority for automation decisions in Section 10.
Decision-Grade Evidence Hierarchy (Use for Build/Buy and Rollout Decisions)
- Independent operator outcomes (required for rollout).
Holdout or controlled LONGSHOT measurements on qualified outcomes (funded activation,first qualified trade,D30 retained traders,spread/depth/fill quality) are mandatory before scale-up. E2 E3 E4 - Policy and channel constraints (required safety gate).
Search/platform/community/compliance rules can invalidate otherwise promising automation and must be checked before rollout. E6 E79 E91 E120 E121 - Capability and operability evidence (conditional, not sufficient alone).
Vendor capability docs, operability artifacts, and pricing establish availability/cost but cannot independently justify production rollout. E78 E81 E82 E92 E93 E94 E95 E108 E109 E110 E111 E112 E113 E114 E115 E116 E117 - Macro discovery-shift indicators (priority hints only).
Macro crawler/referral shifts are useful for experiment prioritization, not direct lift proof. E83 E84
Non-Decision Evidence (Do Not Use Alone)
- Social engagement counts (likes, followers, retweets, comments).
- Community vote totals (HN points, Reddit upvotes).
- Valuation/funding headlines without operating reliability evidence.
- Launch-announcement posts without longitudinal operator outcomes.
Tool Evaluation Protocol (Quarterly)
- Outcome baseline check: define the independent qualified outcomes required for rollout and current baseline values.
- Policy check: ensure channel/content automation remains within current platform/search/community policy.
- Capability and operability check: confirm feature availability, maturity, and fallback/exit path.
- Cost-floor check: compare entry pricing and likely scale cost versus internal build cost.
- Pilot check: run a 14-day test tied to baseline KPIs, then decide
adopt,watch, orreject.
March 2026 Reference Pack Used
- Independent outcome criteria: E2 E3 E4
- Platform/AI capability signals (availability only): E78 E81 E82 E92 E93 E94 E95
- Operator-grade build-vs-buy references (operability/cost): E108 E109 E110 E111 E112 E113 E114 E115 E116 E117
- Discovery/policy controls: E6 E79 E83 E84 E91 E120 E121
March 4, 2026 Rerun: Sales + TOF Tool fp-check (Twitter/X, Telegram, Discord)
Purpose: this is a maturity + cost + operability filter. It is not evidence that these tools improve LONGSHOT outcomes without holdouts.
Verification gates used (adapted for tool adoption):
- Fit: solves a complex workflow (not a trivial build).
- Traction: at least one credible signal of real usage (OSS stars, reviews, or sustained community usage).
- Cost floor: public pricing or low-risk entry tier; no enterprise-only contract as the default path.
- Reachability: supports Twitter/X and/or Telegram/Discord directly, or fits as glue (CRM/enrichment).
- Policy safety: can be used without pushing LONGSHOT into spammy automation.
- Exit path: you can replace it with in-house implementation if needed (API/export or OSS).
TRUE POSITIVES (Use When Triggered)
TOOL #1 TRUE POSITIVE — Attio (CRM + MCP-driven automation)
Fit PASS: CRM + pipeline operations are high-effort to build correctly.
Traction PASS: established product footprint; public pricing.
Cost floor PASS: free entry tier; paid tiers are transparent.
Reachability PASS: sits downstream of all TOF channels.
Policy safety PASS: does not imply any outbound-channel spam automation by itself.
Exit path PASS: API/export exists; avoid deep lock-in by keeping sources-of-truth in your data warehouse.
Evidence: E128 E129
TOOL #2 TRUE POSITIVE — Typefully (Twitter/X content pipeline + MCP server)
Fit PASS: MCP lets you wire drafting/scheduling into an engineering-first agent workflow; X scheduling + campaign UX is non-trivial to rebuild (and X API constraints matter).
Traction PASS: meaningful creator adoption signals; ongoing product velocity.
Cost floor PASS: paid plan required for advanced features (Auto-DMs is explicitly tied to a paid plan); pricing is public.
Reachability PASS: first-class for Twitter/X (primary TOF).
Policy safety CONDITIONAL: safe if used for drafting + scheduling + consented DM flows; unsafe if used for bulk spam.
Exit path PASS: keep content + metrics mirrored internally; avoid tool-only storage.
Evidence: E123 E124 E125 E151
TOOL #3 TRUE POSITIVE — Hypefury (Twitter/X scheduling + AI drafting)
Fit PASS: scheduling/analytics/DM workflow has meaningful surface area; cheaper than building UI + safely handling X API constraints.
Traction PASS: large creator-userbase signal; widely referenced in creator ecosystems.
Cost floor PASS: pricing is public with low entry tiers.
Reachability PASS: first-class for Twitter/X.
Policy safety CONDITIONAL: same spam constraints as all X automation; keep human review + explicit consent.
Exit path PASS: mirror posts/metrics to internal store.
Evidence: E126
TOOL #4 TRUE POSITIVE — Botpress (AI agent platform for Telegram + custom Discord bot)
Fit PASS: durable bot UX, tools, and guardrails are non-trivial; a platform can reduce glue-code and iteration time.
Traction PASS: OSS repo + sustained maintenance signal.
Cost floor PASS: public pricing and OSS foundation.
Reachability PASS: Telegram is supported; Discord requires integration work (fine for strong engineers).
Policy safety PASS: can be run as human-in-loop triage/support rather than unsolicited outbound.
Exit path PASS: OSS + you can port logic to a custom bot later.
Evidence: E130 E131 E132 E133
TOOL #5 TRUE POSITIVE — Clay (enrichment + research automation, used surgically)
Fit PASS: multi-source enrichment + structured research is complex to build and maintain.
Traction PASS: meaningful third-party review footprint.
Cost floor PASS: public pricing; start small and treat as a variable-cost experiment.
Reachability PASS: downstream of TOF capture; helps convert warm social leads into reachable contacts.
Policy safety PASS: no channel policy risk if used for enrichment only.
Exit path PASS: exported outputs + internal enrichment fallbacks.
Evidence: E140 E141
TOOL #6 TRUE POSITIVE — Instantly / Smartlead (email follow-up infra, if you choose outbound)
Fit PASS: deliverability + sequencing is non-trivial to build safely.
Traction PASS: large public review footprint (Instantly) and non-zero footprint (Smartlead).
Cost floor PASS: public pricing.
Reachability CONDITIONAL: only relevant if you turn social/community interest into compliant email follow-up.
Policy safety CONDITIONAL: depends on list quality + consent + compliance; not a default growth lever.
Exit path PASS: providers can be swapped; keep sending domains + prospect lists internal.
Evidence: E145 E146 E147 E148
WATCHLIST (Promising Fit, But Evidence Not Yet Decision-Grade)
TOOL #7 WATCH — Mava (Discord/Telegram support + AI)
Fit PASS: it is channel-aligned (Discord + Telegram) and solves real workflow friction (tickets + support routing).
Traction UNCLEAR: evidence here is mostly vendor-claimed; treat as a 14-day pilot only.
Cost floor PASS: public pricing with a free tier.
Reachability PASS: Discord + Telegram are first-class.
Policy safety PASS: support-first automation is usually safer than outbound automation.
Exit path PASS: keep transcripts + tags exportable; be ready to switch to a self-hosted helpdesk if needed.
Evidence: E134 E135 E136
TOOL #8 WATCH — ManyChat (Telegram marketing automation + AI)
Fit PASS: non-engineers can iterate on Telegram onboarding flows and lead capture without constant dev support.
Traction PASS: long-lived product with substantial public review footprint.
Cost floor PASS: public pricing and low-risk entry tiers.
Reachability PASS: Telegram is first-class.
Policy safety CONDITIONAL: keep automation opt-in and avoid bulk DM spam behaviors.
Exit path PASS: flows can be re-implemented as an in-house Telegram bot when stable.
Evidence: E137 E138 E139
TOOL #9 WATCH — Apollo (prospect database + sequencing)
Fit PASS: contact databases are not realistically “build it yourself”.
Traction PASS: large public review footprint.
Cost floor PASS: public pricing with a free tier.
Reachability CONDITIONAL: only relevant if LONGSHOT chooses outbound email as an explicit motion.
Policy safety CONDITIONAL: database + sequencing can drift into spam without strict list-quality and compliance gates.
Exit path PASS: export lists + keep ICP logic internal.
Evidence: E143 E144
FALSE POSITIVES (Reject As Default)
TOOL #10 FALSE POSITIVE — “AI SDR replacement agents” as default TOF motion Gate 3 (Cost floor) FAIL: many are priced for enterprise budgets and assume high-volume outbound. Gate 5 (Policy safety) FAIL: incentives push toward spammy automation; risk of channel bans and reputation damage. Evidence: E149 E150