6.2Express troubleshooting
The essentials in 30 seconds
- Reflex: one symptom = likely causes = one targeted action — diagnose before relaunching.
- Save: filtering early, targeting the trigger and calling AI only when needed keep quotas under control.
- Secure: every failure must be visible (retry, alert) and no secret stays in plain text.
Lina thread — the playbook taped next to the screen
Lina keeps this playbook next to her screen. Verdict after one year: three rows served her most — the scenario that does not trigger, the duplicates created in the base, and the workflow that failed without her seeing it. For each, she ended up applying the sequence (check the source, search before creating, add retry + alert) instead of relaunching at random.
Observed symptom, verified causes, actions in order: the reflex holds in three steps.
Diagnose before relaunching
A no-code problem is rarely solved at random: each symptom has typical causes and a targeted action. The table below groups the most frequent situations, from data to automation and AI.
| Symptom | Likely causes | Actions |
|---|---|---|
| My scenario does not trigger | trigger misconfigured (wrong account/table); scenario not activated; polling delay; filter placed too early | check the connected account and the source; activate the scenario; wait (polling) or switch to webhook; consult the run history |
| Duplicates are created in my base | no search before creation; trigger relaunching itself; data duplicated instead of linked | add a “search then create/update” step; link tables (linked records) instead of recopying |
| My operations quota explodes | filter missing or too late; trigger too frequent; AI calls on every event | filter early; target the trigger; call AI only when needed; group scenarios |
| AI returns unusable answers | vague prompt; no imposed format; data too bulky | specify the instruction; ask for JSON; send only the useful text; test the prompt |
| AI invents things (hallucinations) | answer from memory with no sources; RAG missing or badly indexed | set up RAG on your documents; ask for cited sources and saying “I don’t know”; human check |
| An AI agent takes an unwanted action | too many tools; fuzzy goal; no human validation | strictly limit tools; bound the goal; human in the loop for sensitive matters; step limit |
| My Glide/Softr app is a mess | underlying base badly structured (text fields, no relations) | re-model the base (types, relations) — the app mirrors the base; then regenerate the screens |
| A workflow failed without me seeing it | no alert; no error handling | add an error handler + alert; retry on transient errors; fallback on critical matters |
| An API key lies in plain text / has leaked | secret inside the scenario or an export; rights too broad | move to encrypted connection/environment variable; regenerate the key immediately; limit its rights |
| Nobody knows how an automation works | undocumented workflow; dependence on a single person (shadow IT) | document (role, trigger, owner); name clearly; plan a recovery procedure |
Ready-to-say script — diagnosing without relaunching at random (30 s)
«My symptom: [idle scenario / duplicates / quota / AI / silent failure]. I check the source and the run history, I identify the likely cause from the playbook, I apply the targeted action and I make failure visible (retry + alert).»
Copy-paste template — express failure sheet
SYMPTOM observed (trigger, duplicates, quota, AI, app, secret): ...
Since when / which scenario / which base: ...
Likely CAUSES (playbook 6.2): ...
CHECK (account, source, run history): ...
ACTIONS (in order): 1) ... 2) ... 3) ...
SECURITY (secret encrypted? rights limited?): ...
DOCUMENTATION (role, trigger, owner): ...VSE / small budget
No formal table: one symptom, the playbook’s likely cause, the action and the owner noted.
SME Algeria
Build local realities into the diagnosis: shared accounts, activity peaks, usual notification channels.
Remote / hybrid
Add immediate written alert to the actions: at a distance, an unnotified failure looks like a system that works.
Intercultural
Spell out labels in writing: what seems obvious (who alerts, which channel) is not obvious to everyone.
48h challenge — apply tomorrow
Spot 1 current symptom (temperamental scenario, duplicates, quota), fill in the express failure sheet and test the 1st action (check + filter early).
- I can link a symptom to its likely causes.
- I filled in (or scheduled) an express failure sheet.
- I know which level to revisit for my case.
Lina’s scenario creates duplicates in the base on every run. Which reflex does the playbook apply?
One recurring symptom is not ten failures: it is a missing reflex to anchor (filter early, search before creating, retry + alert, document).