E-commerce
July 1, 2026
"shipping", "my package", "order 8842", "refund": the customer writes just like in real life. A strict bot replies "I didn't understand" and opens a chattyp_ ticket.
An e-commerce client language AI chatbot tolerates mistakes (fuzzy match), maps synonyms and abbreviations, normalizes slang, and gently confirms if the guess is uncertain.
This guide #890 covers intents bot_lang_*, flow CUSTLANGbot CLB-1 to CLB-8, and KPI lang_bot. Bot pair of the CHATTYP playbook (#889). Usecase: tolerant linguistic understanding on the widget side.
Summary
Why adapt the bot to the customer's language?
Customers don't write the way the helper hub does. FUZZY-TOLERANT-BOT and SYNONYM-GROUND-BOT reduce tickets #889 and complete synonym_map #880 without confusing it with mixed multilingual #891.
What linguistic tolerance solves
Typos: livraision → livraison map
Synonyms: colis paquet → commande map
Abbreviations: cmd sav pb → intent map
Slang: normalization to canonical term map
Fewer tickets: decrease in chattyp_spelling_error
DTC Retail Example
DTC Fashion, fuzzy + optimized synonym_map #889. lang_bot_chattyp_deflect +34%, lang_bot_fuzzy_hit_rate 78% in 6 weeks.
CUSTLANGbot #890 vs CHATTYP #889, MISUND #880, CHATMIS #879 and mixed #891
Five language contents, five distinct roles.
Quick matrix
#890 CUSTLANGbot: fuzzy synonyms abbreviations business glossary bot-side
CHATTYP #889: agents manage tickets formatting errors CT-5
MISUND #880: loop intents incorrectly routed general
CHATMIS #879: off-topic distinct language input
#891 mixed: code-switching multilingual hybrid conversations
Audio #958: fuzzy promo code distinct audio
#890 = tolerate misspelled FR. #891 = customer mixes multiple languages.
Which bot_lang_* intents should be configured?
Eight linguistic intents mapped CHATTYP-MAP #889.
Eight bot_lang intents
bot_lang_fuzzy_match: approximate spelling correction map
bot_lang_synonym_resolve: parcel package → order map
bot_lang_abbrev_expand: cmd sav pb → full term map
bot_lang_slang_normalize: informal speech → canonical map
bot_lang_autocorrect_recover: mobile autocorrect patterns map
bot_lang_confirm_guess: confirm guessed term map
bot_lang_glossary_metier: retail industry expressions map
bot_lang_chattyp_feed: consume LOG #889 enrich map
Each match logs original_term resolved_intent confidence.
How to consume CHATTYP-MAP #889?
The bot reads CHATTYP-MAP #889 + lang fields: fuzzy_threshold, synonym_map, abbrev_map, glossary_metier, chattyp_feed_priority, confirm_guess_threshold.
Customer language guardrails
FUZZY-TOLERANT-BOT: Levenshtein match or equivalent below threshold
SYNONYM-GROUND-BOT: synonym_map return refund parcel
ABBREV-EXPAND-BOT: order after-sales-service problem resolved before intent
NO-SPELLING-SHAMING-BOT: never publicly correct customer spelling
CONFIRM-GUESS-BOT: confirm_guess if medium confidence
GLOSSARY-METIER-BOT: documented DTC retail terms
CHATTYP-FEED-LOOP-BOT: weekly chattyp_feed enriches maps
CUSTLANGBOT-SUP policy in six rules
Six rules for responsible language comprehension.
FUZZY-TOLERANT-BOT: tolerate common spelling mistakes in customer service intents
SYNONYM-GROUND-BOT: synonym_map maintained since LOG #889
ABBREV-EXPAND-BOT: top client abbreviations resolved
CONFIRM-GUESS-BOT: confirm if fuzzy matches are ambiguous between two intents
NO-SPELLING-SHAMING-BOT: no "you wrote it wrong" messages
CHATTYP-FEED-LOOP-BOT: review maps within 48 hours after LOG #889
Flow CUSTLANGbot CLB-1 to CLB-8
Eight-step flow: incoming message normalization fuzzy intent response log feed.
CLB-1 Ingest message: tokenize raw customer text
CLB-2 Abbrev expand: ABBREV-EXPAND cmd sav pb
CLB-3 Fuzzy + synonym: FUZZY-TOLERANT + SYNONYM-GROUND
CLB-4 Slang normalize: glossary_metier if industry phrase
CLB-5 Confidence gate: high → intent; medium → CONFIRM-GUESS
CLB-6 Respond: customer service response from resolved intent
CLB-7 Chattyp feed: failures → chattyp_feed backlog #889
CLB-8 Log: original_term resolved_intent fuzzy_hit chattyp_deflect
TPL-LANGbot-CONFIRM Example
“Are you talking about {{terme_devinu00e9}}? [yes/no] CONFIRM-GUESS-BOT.”
TPL-LANGbot and touchpoint templates
Four short templates for linguistic tolerance embedded.
TPL-LANGbot-CONFIRM
[confirm_guess_copy map.] CONFIRM-GUESS-BOT. No business response before confirmation.
TPL-LANGbot-RESOLVED
[réponse_intent map.] Silent fuzzy resolution. NO-SPELLING-SHAMING.
TPL-LANGbot-CLARIFY
[clarify_copy map.] If fuzzy ambiguous between two close intents.
TPL-LANGbot-GLOSSARY
[glossary_copy map.] Business expression explained, then intent.
Touchpoints
1-2 letter typo: direct fuzzy_match
Colis paquet envoi (Package parcel shipment): synonym_resolve order
cmd + number: abbrev_expand + lookup order
LOG CHATTYP #889: chattyp_feed enriches maps
Edge cases and reroutes
Five cases outside the standard flow.
Mixed multilingual message: #891 mixed not fuzzy FR only
Ambiguous fuzzy delivery return: CONFIRM-GUESS not blind guess
Bot off-topic despite match: #879 CHATMIS
Misspelled promo code: #958 fuzzy code
Offensive slang: distinct moderation not synonym_map
Essential bot language KPIs
Five CUSTLANGbot steering metrics and correlation #889.
lang_bot_fuzzy_hit_rate: % typos resolved without a chattyp_ ticket
lang_bot_synonym_resolve_rate: synonyms mapped correctly
lang_bot_chattyp_deflect: conversations resolved without ticket #889
lang_bot_confirm_save_rate: % confirm_guess avoiding bad intent
lang_bot_map_growth_weekly: new LOG #889 terms integrated
Target: fuzzy_hit_rate rising and chattyp_deflect correlated with enriched maps.
CUSTLANGbot anti-patterns
Five common linguistic tolerance errors.
Requiring perfect spelling: FUZZY-TOLERANT mandatory for customer service
Correcting the customer: strict NO-SPELLING-SHAMING
Static synonym_map: weekly CHATTYP-FEED-LOOP #889
Too broad fuzzy: CONFIRM-GUESS if delivery return is ambiguous
Confusing #891: mixed multilingual reroute, not FR abbreviation
CUSTLANGbot with Qstomy
Qstomy on Shopify: fuzzy match, editable synonym_map, CHATTYP-MAP feed #889, professional glossary, KPI lang_bot dashboard.
Scenario: retail DTC, 15 chattyp_ expressions/month. Maps enrichment from LOG #889. lang_bot_chattyp_deflect +34%, lang_bot_fuzzy_hit_rate 78% in 6 weeks.
Explore AI support and request a demo.
Checklist, FAQ and going further
CUSTLANGbot Checklist (8 steps)
Sync CHATTYP-MAP #889: feed spelling synonym abbreviation
Policy CUSTLANGBOT-SUP: 6 FUZZY SYNONYM NO-SHAMING rules
8 intents bot_lang_*: flow CLB-1 to CLB-8
4 templates TPL-LANGbot-*: CONFIRM RESOLVED CLARIFY GLOSSARY
synonym_map SAV: return refund parcel order delivery
abbrev_map top 20: order cs pb deployed
Red team typos: livraision rembousment fuzzy test
KPI Dashboard: lang_bot_* section 9 + delta chattyp_
FAQ
Difference #889?
#889 = agents process typo tickets. #890 = bot tolerate and map widget-side.
Difference #880?
#880 = general poorly routed intents loop. #890 = fuzzy synonyms linguistic abbreviations.
Difference #891?
#890 = misspelled FR. #891 = mix of multiple conversational languages.
Fuzzy too permissive?
CONFIRM-GUESS if ambiguous. Adjust fuzzy_threshold per intent.
Going further
This week: deploy synonym_map top 10, enable FUZZY-TOLERANT test threshold, sync LOG #889 weekly, measure lang_bot_chattyp_deflect.
Enzo
July 1, 2026
