Automated Analysis Disclaimer
Version: 2026.2 Effective Date: 27 May 2026 Status: Active
This Automated Analysis Disclaimer forms part of the Outpost Terms of Service. By accepting the Terms of Service, you also accept this Disclaimer.
The Services generate outputs using Automated Analysis. Automated Analysis is powerful but fallible. It can confidently produce text and numbers that are wrong, out of date, fabricated, inconsistent, or incomplete. If you treat those outputs as finished, authoritative, or decision-ready, you can cause real-world harm to yourself, your business, your workers, your counterparties, and third parties - and you can suffer legal, financial, regulatory, and reputational consequences.
If you do not accept this Disclaimer, you must not use the Services.
1.1 "Automated Analysis" has the meaning given in the Terms of Service. It includes any output generated by automated software systems - including large language models, other machine-learning systems, statistical models, rules engines, heuristics, optimisation routines, and combinations of the above - whether presently deployed or deployed after the Effective Date.
1.2 Automated Analysis is carried out by software. A human subject-matter expert does not review each output before it appears to you. The software is probabilistic: the same input may produce different outputs on different runs.
1.3 This Disclaimer applies to every Automated Analysis output, regardless of module, format (text, numbers, images, tables, files, code, spreadsheets), presentation (screen, export, email, API, notification, PDF), or delivery channel.
2.1 Every Automated Analysis output is a draft. You must treat every output as a starting point produced by software, not as:
(a) a finished document; (b) an established fact; (c) a decision; (d) a certification, sign-off, or approval; (e) professional advice of any kind; or (f) an output of a human expert.
2.2 Nothing in the Services creates a professional, client, fiduciary, advisory, consulting, or employment relationship between you and the Operator. The Operator is a software provider. It does not hold itself out as possessing the expertise of any licensed practitioner.
2.3 Automated Analysis outputs are not decisions that produce legal or similarly significant effects on individuals on our behalf. You, the User, remain the decision-maker in every case and are solely responsible for any decision you make.
3.1 Automated Analysis outputs can and do contain errors, including the following categories. This list is non-exhaustive.
(a) Hallucinations and fabrication. The models can produce confident, fluent, well-structured text and numbers that are entirely false. Cited references may not exist. Quoted passages may not appear in the source. Named people, companies, standards, clauses, and authorities may be invented.
(b) Factual errors. The output may assert things that are simply wrong, including about your project, your contract, your people, your site, applicable law, applicable Standards, historical events, prices, rates, or numbers.
(c) Arithmetic and numerical errors. Totals, subtotals, quantities, rates, durations, percentages, multipliers, and unit conversions can be miscalculated, inconsistent within the same document, or inconsistent across runs.
(d) Document misreading. Ingested PDFs, scans, emails, images, WhatsApp messages, and drawings can be misread or mis-OCR'd. The output may rely on a misreading without flagging it.
(e) Omission and missed clauses. The output may miss clauses, obligations, deadlines, conditions precedent, exclusions, qualifications, annexures, or attachments in the source material. Missing a single clause in a tender, contract, or notice can be fatal.
(f) Stale reference data. The models are trained on data with a cutoff date. Standards, codes, legislation, rates, and market conditions may have changed since the cutoff. The Services may incorporate additional reference data, but that data is also not guaranteed current.
(g) Non-determinism. Running the same input twice can produce different outputs. Do not assume a favourable output is reproducible. Do not assume an unfavourable output cannot recur.
(h) Bias and systematic error. The models inherit biases from their training data and from any reference data used by the Services. Biases can affect how people, companies, places, prices, or risks are characterised.
(i) Prompt-injection and data-poisoning. Hostile content inside uploaded documents, emails, or integration data can influence the output in ways you do not expect. This is a live risk in any Automated Analysis system.
(j) Integration ingestion failure. Inbound integrations can miss emails, misparse tenders, duplicate WhatsApp messages, time-out on large files, drop attachments, process the wrong version, import out-of-order, or fall behind the real world. The Services cannot act on data that did not arrive, and cannot always tell you that data is missing.
(k) Misclassification. Outputs that classify, triage, score, or prioritise (for example, bid/no-bid recommendations, risk scores, tender triage) can misclassify. A "no-bid" output may be wrong. A "low-risk" output may be wrong.
(l) Unit, currency, jurisdiction, and timezone mix-ups. Outputs can mix imperial and metric, use the wrong currency, apply the wrong jurisdiction's law or standards, or apply the wrong timezone to a deadline.
(m) Chaining and amplification. When one Automated Analysis output is used as input for another (for example, programme → cost → claim), errors compound. An error early in the chain can silently propagate.
(n) Inconsistency across modules. Different modules of the Services may produce outputs that are inconsistent with each other. Reconciling them is your responsibility.
(o) Confident tone. Outputs are usually written in a confident, authoritative tone. Tone is not evidence of accuracy. A confidently-worded output can be completely wrong.
3.2 The Operator does not and cannot warrant that any Automated Analysis output is free of any of the failure modes in clause 3.1, or of any other failure mode.
4.1 Before you rely on any Automated Analysis output, you must independently review, verify, and where necessary correct the output using your own judgement, expertise, and - where required - the input of qualified and licensed professionals.
4.2 "Rely on" includes, without limitation, each of the following actions:
sending the output to any third party;
signing the output;
issuing, serving, or submitting the output to any counterparty, principal, subcontractor, adjudicator, court, regulator, or authority;
lodging, filing, or publishing the output;
acting on the output;
quoting or pricing based on the output;
invoicing against the output or paying against it;
building to the output or using it for construction, procurement, or planning;
scheduling around the output;
making a decision about any person, project, or transaction based on the output;
representing to any third party that the output is accurate, complete, reviewed, approved, or fit for purpose;
incorporating the output into another document or system; or
deleting or ignoring the source material in reliance on the output.
4.3 Human review must be performed by a person with the training, experience, authority, and (where required) licensing or registration to evaluate the output in its context. For outputs within the categories listed in clause 2 of the Construction Industry Disclaimer, the reviewer must meet the qualifications in clause 3 of that Disclaimer.
4.4 You must not disable, remove, obscure, alter, or misrepresent any "DRAFT", "UNVERIFIED", or similar label, banner, watermark, or warning that the Services apply to an output. You must not represent to any third party that an output has been reviewed, approved, or certified when it has not.
4.5 Operator product commitments. In support of this clause 4, the Operator commits that:
(a) every file exported from the Services (including PDF, Word, Excel, PowerPoint, CSV, and ZIP exports) carries a visible "Outpost draft - requires review" marker, together with the version and timestamp of the output; (b) the Services require a per-export acknowledgement from the User at the moment an output within any of the high-risk categories listed in clause 2 of the Construction Industry Disclaimer is exported, submitted, served, or otherwise released from the Services. That acknowledgement is in addition to, and does not replace, your general acceptance of this Disclaimer; and (c) the Services will, where technically feasible, flag outputs that exhibit elevated uncertainty (for example, where confidence indicators, calibration metrics, or cross-check signals are unfavourable) so that you may direct additional human review to those outputs.
The Operator may modify the technical implementation of these commitments from time to time, provided the substance of each commitment is maintained.
5.1 Automated Analysis outputs are not:
(a) legal advice; (b) engineering advice; (c) architectural, surveying, or building-practitioner advice; (d) work health and safety advice; (e) financial, accounting, tax, or audit advice; (f) insurance or risk-management advice; (g) planning, environmental, heritage, or cultural-heritage advice; (h) investment, valuation, or market-forecasting advice; (i) procurement or probity advice; or (j) any other form of professional, regulated, or licensed advice.
5.2 Any statement in an Automated Analysis output that reads like advice must be treated as a draft suggestion subject to review by a qualified and, where required, licensed professional in the relevant field.
5.3 Market intelligence, benchmarking, pricing, cost-estimating, forecasting, and competitor outputs are informational only. They are not financial advice, investment advice, valuation advice, or tax advice. Reliance on them in any financial, commercial, or investment decision is at your sole risk.
6.1 The Services can only work with data that has actually been received. If a Third-Party Integration fails to deliver data (because of an outage, quota, permission change, account change, spam filter, API change, format change, size limit, timeout, retry exhaustion, or any other reason), the Services will process without that data and may not know it is missing.
6.2 You are responsible for ensuring that:
(a) all critical data has actually been ingested; (b) nothing important has been missed, duplicated, or misrouted; (c) inbound emails, tenders, WhatsApp messages, drive files, and integration events have arrived and been processed correctly; and (d) you maintain independent copies and backups of anything you cannot afford to lose.
6.3 The Operator is not responsible for any loss arising from failure of any Third-Party Integration, from failure of the Services to ingest particular data, or from your reliance on an Automated Analysis output that was produced without the benefit of missing data.
7.1 Automated Analysis outputs may include statements about third parties - subcontractors, suppliers, competitors, principals, individuals - drawn from Automated Analysis of User Content and public sources. Statements about third parties can be wrong, misleading, or defamatory.
7.2 If you repeat, publish, forward, quote, or otherwise communicate an Automated Analysis output externally, you (not the Operator) are responsible for:
(a) verifying the factual accuracy of any statement about any person or entity; (b) ensuring the statement is not misleading or deceptive, or likely to mislead or deceive, contrary to section 18 of the Australian Consumer Law; (c) ensuring the statement is not defamatory under the Defamation Act 2005 (model) or any equivalent law; (d) ensuring the statement does not breach any passing-off, trade-mark, confidentiality, or privacy right; (e) ensuring the statement does not breach any contractual duty of good faith or cooperation you owe to the third party; and (f) ensuring the statement complies with any non-disparagement, confidentiality, or media clause in any contract to which you are a party.
7.3 The Operator accepts no liability for any claim against you arising from your external use of an Automated Analysis output. This clause is cumulative with clause 14 of the Construction Industry Disclaimer.
8.1 You must not use Automated Analysis outputs as the sole basis for any decision that produces a legal effect, or a similarly significant effect, on an individual - including any decision about:
(a) engagement, non-engagement, termination, or disciplinary action in respect of a worker, subcontractor, or supplier; (b) whether to admit a person to site, issue an induction, or refuse entry; (c) incident, injury, or safety findings about a person; (d) creditworthiness, solvency, or financial standing of a person or entity; (e) allegations of misconduct, fraud, or dishonesty against a person; or (f) any other determination that, if wrong, would materially affect a person's rights, livelihood, safety, or reputation.
8.2 Decisions of that kind require human review, human judgement, and, where required, independent verification. This clause is included to align with Australian Privacy Principle obligations and the Privacy Act 1988 (Cth) as amended from time to time, and with equivalent overseas law (including GDPR Article 22 in the European Economic Area and UK GDPR Article 22 in the United Kingdom).
9.1 The Operator may at any time change:
(a) the underlying models used by the Services; (b) the prompts, rules, heuristics, and instructions used; (c) the reference data, embeddings, indices, and knowledge bases used; (d) the providers of Automated Analysis; (e) the scope, cost, latency, and quality of Automated Analysis outputs; and (f) the credit consumption rate of Automated Analysis operations.
9.2 Automated Analysis outputs are not guaranteed to be reproducible across versions, models, providers, or runs. An output produced today may not match an output produced tomorrow on the same input. This is a feature of how machine-learning systems work, not a defect.
9.3 Credit consumption does not correlate with quality. An output that consumed more credits is not necessarily more accurate, more reliable, or more fit for purpose than one that consumed fewer.
10.1 The Operator's rights in respect of User Content submitted for Automated Analysis, and the Operator's ownership of Aggregated Data derived from Automated Analysis outputs, are set out in clause 8 of the Terms of Service and the Privacy Policy.
10.2 Consistent with clause 8.8 of the Terms of Service, the Operator will not submit raw, identifiable User Content to third-party foundation-model training pipelines without your express opt-in consent.
10.3 Automated Analysis outputs may reflect patterns learned from Aggregated Data. That is intended and authorised under clause 8 of the Terms of Service.
11.1 The Operator strongly recommends that you keep records of:
(a) each Automated Analysis output you rely on; (b) the human review performed before reliance; (c) the qualifications and identity of the reviewer; (d) any corrections made to the output before reliance; (e) the date, time, and version of the output; and (f) the decisions taken based on the output.
11.2 Records of this kind help you demonstrate compliance with your professional, statutory, insurance, and contractual obligations, and help you defend a claim arising from reliance on an output. The audit-trail module of the Services may assist with this record-keeping, but is not a substitute for your own records.
12.1 You expressly assume the risk of using Automated Analysis. You acknowledge that:
(a) Automated Analysis outputs can and do contain each of the failure modes listed in clause 3, and other failure modes not listed; (b) you have been specifically warned of the need for independent human review of every output before reliance; (c) you are in a better position than the Operator to assess the accuracy, completeness, and appropriateness of any output in the context of your specific project, contract, site, workforce, jurisdiction, and counterparties; (d) the commercial bargain reflected in the credit-metered pricing is predicated on this allocation of responsibility; (e) the Operator does not hold itself out as possessing the expertise of any licensed practitioner and does not voluntarily assume a duty of care in tort to you or to any third party to produce accurate, complete, context-specific, or reproducible Automated Analysis outputs, or to warn you that a particular output is wrong, stale, missing data, or unsuitable for a particular purpose; and (f) you will not assert in any proceeding that any Automated Analysis output was reasonably relied on without independent review.
12.2 Limitation of any duty found to exist. If, contrary to clause 12.1(e), a duty of care in tort is nevertheless found to exist on the facts of a particular matter:
(a) the duty is limited in scope to the exercise of reasonable care in the design, operation, and maintenance of the Services as software, and does not extend to the substantive accuracy, completeness, or fitness for purpose of any individual output; (b) the standard of care is informed by this Disclaimer, by the Construction Industry Disclaimer, by your assumption of risk, and by your mandatory human-review obligation; and (c) any liability arising under any such duty is subject to the liability cap, exclusions, and carve-outs in clause 15 of the Terms of Service.
12.3 Contributory conduct. You acknowledge that any failure by you to perform the human review required by clause 4 constitutes contributory conduct that, to the maximum extent permitted by law, reduces or extinguishes any liability of the Operator in connection with the relevant output.
12.4 This clause is cumulative with the assumption-of-risk, duty-of-care-limitation, and non-reliance clauses in the Construction Industry Disclaimer and the Terms of Service.
13.1 The legal environment for machine-learning-assisted outputs is evolving, including through the voluntary AI Safety Standard, potential Australian mandatory guardrails for high-risk AI, Privacy Act reform tranches, and equivalent developments overseas.
13.2 Subject to the Terms of Service and any non-excludable law, the Operator may change, restrict, add disclosures to, pause, or discontinue any Automated Analysis feature in order to:
(a) comply with any law, regulation, guideline, standard, code of practice, or authority direction now or in the future applicable to automated analysis, machine learning, or high-risk AI; (b) comply with a request or direction from a provider, sub-processor, or regulator; (c) respond to a legal claim, threatened claim, or identified risk; or (d) protect users, third parties, or the public.
13.3 Changes made under this clause are deemed immaterial for the purposes of clause 25 of the Terms of Service where they are required by law, regulation, or authority direction, and do not trigger re-acceptance, except where re-acceptance is required to inform you of a material change in how Automated Analysis is performed.
14.1 In the Services' user-facing interface, the Operator refers to its machine-learning features using the term "Automated Analysis" or "automated tools", and avoids the term "AI" and the names of underlying model providers. This is a branding choice. Nothing in the Services' branding is intended to obscure the reality - disclosed in this Disclaimer - that Automated Analysis is carried out using machine-learning systems including large language models.
15.1 To the maximum extent permitted by law, the Operator disclaims liability for any loss, damage, claim, penalty, fine, injury, or cost arising out of or in connection with your use of, or reliance on, any Automated Analysis output, including in each of the loss categories listed in clause 19 of the Construction Industry Disclaimer and clause 15 of the Terms of Service.
15.2 Without limiting clause 15.1, the Operator disclaims all liability for any loss arising from any failure mode listed in clause 3, whether or not that failure mode was known to the Operator at the time the output was generated.
15.3 This clause is subject to clause 15.3 of the Terms of Service (non-excludable carve-outs) and clause 14 of the Terms of Service (consumer guarantees).
16.1 This Disclaimer survives termination of the Terms of Service to the extent that any Automated Analysis output remains in use or capable of being relied on.
16.2 This Disclaimer is cumulative with the Terms of Service, the Privacy Policy, and the Construction Industry Disclaimer. Where two provisions overlap, the one more protective of the Operator applies, to the extent permitted by law.
Capitalised terms used and not defined here have the meaning given in the Terms of Service. "Third-Party Integration" has the meaning given in the Terms of Service.
By accepting this Disclaimer, you acknowledge that:
End of Automated Analysis Disclaimer.